Next Page: 10000

          Comment on How to Get Started with Deep Learning for Time Series Forecasting (7-Day Mini-Course) by Jason Brownlee      Cache   Translate Page      
Thanks for sharing.
          Ingram Micro Signs Distribution Agreement with NVIDIA for the META region      Cache   Translate Page      

Ingram Micro Inc., one of the largest value-added distributors in the Middle East, Turkey and Africa region, today announced a start of the distribution relationship with NVIDIA, a world leader in artificial intelligence computing technology. With this agreement, Ingram Micro is authorized to sell and promote NVIDIA's Deep Learning, GPU Virtualization and High Performance Computing solutions in the GCC countries (namely, Qatar, Saudi Arabia, United Arab Emirates), Turkey and South Africa....

Read the full story at https://www.webwire.com/ViewPressRel.asp?aId=229825


          Deep learning for numerical analysis explained      Cache   Translate Page      

Deep learning (DL) is a subset of neural networks, which have been around since the 1960’s. Computing resources and the need for a lot of data during training were the crippling factor for neural networks. But with the growing availability of computing resources such as multi-core machines, graphics processing units [...]

Deep learning for numerical analysis explained was published on SAS Users.


          Deep Learning-Experte/Expertin - Bosch Gruppe - Vaihingen an der Enz      Cache   Translate Page      
Standort Stuttgart-Vaihingen Arbeitsbereiche Informationstechnologie Einstieg als Berufserfahrene/-r Startdatum Nach Vereinbarung Arbeitszeit Vollzeit und...
Gefunden bei Bosch Gruppe - Fri, 07 Sep 2018 12:02:46 GMT - Zeige alle Vaihingen an der Enz Jobs
          Deep Learning Expert - Bosch Gruppe - Vaihingen an der Enz      Cache   Translate Page      
Standort Stuttgart-Vaihingen Arbeitsbereiche Informationstechnologie Einstieg als Berufserfahrene/-r Startdatum Nach Vereinbarung Arbeitszeit Vollzeit und...
Gefunden bei Bosch Gruppe - Fri, 07 Sep 2018 12:02:45 GMT - Zeige alle Vaihingen an der Enz Jobs
          Deep Learning for Machine Vision      Cache   Translate Page      
Relatively obscure a few short years ago, Deep Learning is ubiquitous today across data-driven applications as diverse as machine vision, ...
          Machine Learning / Algorithim Developer - TECHNICA CORPORATION - Dulles, VA      Cache   Translate Page      
Job Description: We are seeking a highly creative software engineer experienced in artificial intelligence and deep learning techniques to design, develop,...
From Technica Corporation - Fri, 05 Oct 2018 10:31:19 GMT - View all Dulles, VA jobs
          Deep Learning-Experte/Expertin - Bosch Gruppe - Vaihingen an der Enz      Cache   Translate Page      
Standort Stuttgart-Vaihingen Arbeitsbereiche Informationstechnologie Einstieg als Berufserfahrene/-r Startdatum Nach Vereinbarung Arbeitszeit Vollzeit und...
Gefunden bei Bosch Gruppe - Fri, 07 Sep 2018 12:02:46 GMT - Zeige alle Vaihingen an der Enz Jobs
          Deep Learning Expert - Bosch Gruppe - Vaihingen an der Enz      Cache   Translate Page      
Standort Stuttgart-Vaihingen Arbeitsbereiche Informationstechnologie Einstieg als Berufserfahrene/-r Startdatum Nach Vereinbarung Arbeitszeit Vollzeit und...
Gefunden bei Bosch Gruppe - Fri, 07 Sep 2018 12:02:45 GMT - Zeige alle Vaihingen an der Enz Jobs
          דרוש מהנדס אלגוריתמי Computational Geometry/3D/ (רצוי למידה עמוקה או למידת מכונה)      Cache   Translate Page      
מהנדס אלגוריתמי Computational Geometry/3D (רצוי למידה עמוקה או למידת מכונה).במשרה מלאה, לא מהביתחובה:נסיון בפיתוח אלגוריתמי Computational Geometry/D3/CADC++ או מטלאב או פייתוןM.Sc. או תואר ראשון עם המון נסיון בפיתוח אלגוריתמים כנ"ל.רצוי:Computer VisionDeep Learning or Machine Learning
          Amazon Rekognition improves the accuracy of image moderation      Cache   Translate Page      

Amazon Rekognition is a deep learning-based image and video analysis service that can identify objects, people, text, scenes, and activities, as well as detect unsafe content. Amazon Rekognition now comes with an improved image moderation model that reduces false positive rates by 40% on average without any reduction in detection rates for truly unsafe content. Lower false positive rates imply lower volumes of flagged images to be reviewed further, leading to higher efficiency of human moderators and more cost savings.


          Une intelligence artificielle traduit un livre de 800 pages en 12 heures      Cache   Translate Page      

“L’apprentissage profond”, alias “Deep learning” dans sa version originale est un livre qui sortira en version française le 18 octobre prochain. Outre son titre, ce sont ses quelque 800 pages qui ont été traduites… par une intelligence artificielle. Il s’agira donc du premier ouvrage traduit par une IA au monde. Il ne pouvait en être […]

L’article Une intelligence artificielle traduit un livre de 800 pages en 12 heures est apparu en premier sur Geeko.


          GStreamer: GStreamer Conference 2018: Talks Abstracts and Speakers Biographies now available      Cache   Translate Page      

The GStreamer Conference team is pleased to announce that talk abstracts and speaker biographies are now available for this year's lineup of talks and speakers, covering again an exciting range of topics!

The GStreamer Conference 2018 will take place on 25-26 October 2018 in Edinburgh (Scotland) just after the Embedded Linux Conference Europe (ELCE).

Details about the conference and how to register can be found on the conference website.

This year's topics and speakers:

Lightning Talks:

  • gst-mfx, gst-msdk and the Intel Media SDK: an update (provisional title)
    Haihao Xiang, Intel
  • Improved flexibility and stability in GStreamer V4L2 support
    Nicolas Dufresne, Collabora
  • GstQTOverlay
    Carlos Aguero, RidgeRun
  • Documenting GStreamer
    Mathieu Duponchelle, Centricular
  • GstCUDA
    Jose Jimenez-Chavarria, RidgeRun
  • GstWebRTCBin in the real world
    Mathieu Duponchelle, Centricular
  • Servo and GStreamer
    Víctor Jáquez, Igalia
  • Interoperability between GStreamer and DirectShow
    Stéphane Cerveau, Fluendo
  • Interoperability between GStreamer and FFMPEG
    Marek Olejnik, Fluendo
  • Encrypted Media Extensions with GStreamer in WebKit
    Xabier Rodríguez Calvar, Igalia
  • DataChannels in GstWebRTC
    Matthew Waters, Centricular
  • Me TV – a journey from C and Xine to Rust and GStreamer, via D
    Russel Winder
  • ...and many more
  • ...
  • Submit your lightning talk now!

Many thanks to our sponsors, Collabora, Pexip, Igalia, Fluendo, Facebook, Centricular and Zeiss, without whom the conference would not be possible in this form. And to Ubicast who will be recording the talks again.

Considering becoming a sponsor? Please check out our sponsor brief.

We hope to see you all in Edinburgh in October! Don't forget to register!


          دیدگاه‌ها برای دانلود آموزش Hands-on Deep Learning with TensorFlow با dev-master      Cache   Translate Page      
با سلام مشکل لینک ها برطرف گردید.
          دیدگاه‌ها برای دانلود آموزش Hands-on Deep Learning with TensorFlow با hamedtalebpoorb      Cache   Translate Page      
سلام ببخشید لینک این دانلود خرابه.لطفا درستش کنید.خیلی ممنون
          Del Google Pixel 2 al Google Pixel 3: todo lo que ha cambiado      Cache   Translate Page      

Del Google Pixel 2 al Google Pixel 3: todo lo que ha cambiado#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

Los Pixel no se presentan en solitario. Es algo que Google nos enseñó hace ya dos generaciones y que permanece imborrable en nuestro recuerdo. Como también es imborrable que tenemos cita con el gigante de Mountain View en el mes de octubre para conocer su nueva propuesta. Hoy, la tercera generación de sus dispositivos, los que reemplazaron a los Nexus y que hoy se plasman en los nuevos Google Pixel 3 y Google Pixel 3 XL.

Como ocurrió el pasado año, la renovación de los Pixel se ha basado en una iteración bastante común, actualizando los componentes principales, aunque también hay novedades. Como el hecho de que la cámara frontal se duplica, algo que curiosamente no ocurre con la trasera. Pero vayamos paso a paso y veamos qué ha ocurrido en este salto generación, la evolución de los Pixel 2 a los Pixel 3. ¿Nos acompañas?

Móviles de 2018, cerebros de 2018

Snapdragon 845#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

El salto lógico en estos nuevos Google Pixel 3 y Pixel 3 XL es el de la actualización de sus procesadores. En la pasada generación, los Pixel 2 se vistieron con el Snapdragon 835, el chip predominante en el catálogo de Qualcomm en aquel momento, pese a que un par de meses después afloraría su sustituto, el Snapdragon 845 que se presenta en esta generación.

El nuevo cerebro supone mantenerse en los 10 nanómetros, pero ganar en potencia bruta y, sobre todo, en capacidad de procesamiento de código de inteligencia artificial. Un chip mejor preparado para ejecutar códigos de machine learning y deep learning, favoritos de Google desde hace tiempo para sus Pixel y para otras apps de la casa, como Google Photos o Google Assistant.

Tampoco hay salto en el equipo de memorias, con 4GB de modelo básico con 64GB y 128GB en el almacenamiento. Como en la generación pasada, aquí tampoco hay ranura para expandir la tarjeta microSD, por lo que será mejor tener claro qué capacidad necesitaremos para no andar después lamentándonos de no poder ampliarla. Como vemos, un líder de catálogo de 2018 con todas las de ley, aunque la RAM tal vez sea algo corta. Pero estando Google detrás, optimizando el sistema, puede no ser un factor diferencial.

Las pantallas crecen y se estiran

Pantalla#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

Si hay un detalle estético que va a marcar la diferencia entre los Pixel 3 y los Pixel 2, ése va a ser el del recorte en la pantalla. El archiconocido 'notch', también llamado ceja y que aparece en esta renovación, aunque únicamente en el modelo XL. El pequeño se mantiene rectangular, aunque también estira al adoptar el aspecto 18:9, algo que ya debió ocurrir en el pasado Google Pixel 2.

El Pixel 3 estira hasta los 18:9 y el Pixel 3 XL va más allá, añadiendo el notch a su diseño de pantalla

Así pues, el Google Pixel 3 pasa de tener 5 pulgadas a 5,5 pulgadas, y estira la resolución FullHD hasta el FullHD+ pero manteniendo la densidad. Eso lo logra con el nuevo aspecto 18:9, que le permite llevar la resolución hasta los 2.160 x 1.080 píxeles, respetando los 460 píxeles por pulgada. Como en ocasiones anteriores, tendremos Gorilla Glass 5 protegiendo la pantalla, AMOLED, contra golpes y arañazos.

En cuanto al Pixel 3 XL, su cambio es el más sensible. El pasado año ya teníamos una pantalla 18:9, y ahora tenemos una 18,5:9. ¿Cómo? Pues añadiendo el ya citado 'notch'. Así que las 6 pulgadas del Pixel 2 XL se convierten ahora en 6,3 pulgadas, y la resolución QHD+ se estira algo más y alcanza los 2.960 x 1.440 píxeles. Más larga, más píxeles, una densidad semejante. Y claro, también tenemos Gorilla Glass 5 sobre este panel P-OLED.

Cámaras únicas en la espalda, y dobles en el frontal

Camaras#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

En un primer momento, resultó extraño que Google apostase únicamente por un sensor trasero para sus teléfonos, pues estaban diseñados para competir contra los mejores, y ya emergían con fuerza las cámaras duales con desenfoque selectivo. Pero Google ofreció este desenfoque con un único ojo, y volvió a hacerlo una generación después. También lo hace en esta tercera pero no en el frontal, pues sobre la pantalla se coloca una tercera lente.

Así, el salto en cámaras traseras se mantiene con una sencilla actualización de sensores pero con características casi gemelas. Los Pixel 3 y Pixel 3 XL repiten con sus 12,2 megapíxeles, con lentes de 27 milímetros y apertura f/1.8, estabilizadas ópticamente, y sus sensores son Dual Pixel y tienen píxeles de 1,4 micrómetros. Eso significa que tendremos vídeo 4K con cámara lenta, y el resto de funciones vendrá del apartado de software del teléfono.

Pero en el frontal la cosa se complica, y se duplica. De 8 megapíxeles en los Pixel 2 hemos pasado a 8 y 8 megapíxeles en los Pixel 3. Dos ojos sobre pantalla en el Pixel 3, y dentro de ella en el Pixel 3 XL por su notch. Dos sensores de ocho megapíxeles con lentes f/2.2 y apertura de 27 milímetros, con píxeles de 1,4 micrones y una gran diferencia, que aquí no hablamos de Dual Pixel. Eso sí, habrá desenfoque frontal, tendremos bokeh para los selfies. Y vídeo FullHD, claro está.

2018 es Android Pie, también los Pixel 3

Android Pie#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

Al menos en el ecosistema propio de Google y no dependiente de él. Si los Pixel 2 llegaron con Oreo corriendo por sus venas, los nuevos Pixel 3 y Pixel 3 XL aterrizan con la nueva versión de su sistema, Android 9.0 Pie, lo que hace que los teléfonos cuenten con todos los avances del sistema, como el modo de bienestar digital que pretende que seamos conscientes de cuánto tiempo dedicamos al smartphone cada día.

También tendremos gestos para navegar por el teléfono, pudiendo realizar transiciones entre aplicaciones deslizando el dedo sobre la pantalla. También se estrenan las llamadas Slices, o porciones de información extra que el sistema extrae de las propias aplicaciones para poder lanzarlas desde el propio buscador del sistema. Por supuesto, no estarán ausentes los avances en gestión de batería, con el brillo inteligente que debe reducir el consumo tanto en primer como en segundo plano.

Algunas características inmutables, o casi

A pesar de que ha habido un salto generacional entre los Pixel 2 y los Pixel 3, algunas características han resistido el salto sin apenas cambiar. Como el marco táctil heredado de HTC, el Active Sense que se perpetúa en esta tercera generación de Pixel. O como el hecho de contar con lector de huellas en la espalda. O la conectividad, con WiFi 5, Bluetooth 5.0, GPS o USB tipo C. Incluso el jack de auriculares, que vuelve a estar ausente.

Los nuevos Pixel 3 y Pixel 3 XL mantienen la resistencia al agua IP67, después del IP53 de su primera generación, y sí hay pequeñas variaciones en las baterías. El Pixel 3 cuenta con una pila interna de 2.915 mAh y el 3 XL sube hasta los 3.430 mAh, parecidas a las de los Pixel 2. Ambas con carga rápida y con carga inalámbrica.

Google Pixel 2 vs Google Pixel 3, las características comparadas

Google Pixel 3 Xl#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

Google Pixel 2

Google Pixel 3

Google Pixel 2 XL

Google Pixel 3 XL

Pantalla

5,0" FHD (1920 x 1080)
AMOLED 16:9

5,5" FHD+ (2160 x 1080)
AMOLED 18:9

6,0" QHD+ (2880 x 1440)
P-OLED 18:9

6,3" QHD+ (2960 x 1440)
P-OLED 18.5:9

Procesador

Snapdragon 835

Snapdragon 845

Snapdragon 835

Snapdragon 845

RAM

4 GB

4 GB

4 GB

4 GB

Almacenamiento

64 / 128 GB

64 / 128 GB

64 / 128 GB

64 / 128 GB

Cámara trasera

12.2 megapíxeles f/1.8
Píxeles de 1.4µm
OIS y EIS
Pixel Core
Vídeo 4K/30fps

12.2 megapíxeles f/1.8
Píxeles de 1.4µm
OIS y EIS
Pixel Core
Vídeo 4K/30fps

12.2 megapíxeles f/1.8
Píxeles de 1.4µm
OIS y EIS
Pixel Core
Vídeo 4K/30fps

12.2 megapíxeles f/1.8
Píxeles de 1.4µm
OIS y EIS
Pixel Core
Vídeo 4K/30fps

Cámara frontal

8 megapíxeles f/2.2
Píxeles de 1.4µm
Vídeo 1080p

8 megapíxeles f/2.2
8 megapíxeles f/1.8
Píxeles de 1.4µm
Vídeo 1080p

8 megapíxeles f/2.2
Píxeles de 1.4µm
Vídeo 1080p

8 megapíxeles f/2.2
8 megapíxeles f/1.8
Píxeles de 1.4µm
Vídeo 1080p

Batería

2.700 mAh
Carga rápida

2.915 mAh
Carga rápida
Carga inalámbrica

3.520 mAh
Carga rápida

3.450 mAh
Carga rápida
Carga inalámbrica

Dimensiones y peso

145.7 x 69.7 x 7.8 mm
143 gramos

145.6 x 68.2 x 7.9 mm
148 gramos

157.9 x 76.7 x 7.9 mm
175 gramos

-

Software

Android 8 Oreo (actualizado a Pie)

Android 9 Pie

Android 8 Oreo (actualizado a Pie)

Android 9 Pie

Otros

Sin jack
NFC
Lector de huellas trasero
Bluetooth 5.0
GPS
USB tipo C
Active Edge
WiFi 5 (ac)
Gorilla Glass 5
Protección IP67

Sin jack
NFC
Lector de huellas trasero
Bluetooth 5.0
GPS
USB tipo C
Active Edge
WiFi 5 (ac)
Gorilla Glass 5
Protección IP68

Sin jack
NFC
Lector de huellas trasero
Bluetooth 5.0
GPS
USB tipo C
Active Edge
WiFi 5 (ac)
Gorilla Glass 5
Protección IP67

Sin jack
NFC
Lector de huellas trasero
Bluetooth 5.0
GPS
USB tipo C
Active Edge
WiFi 5 (ac)
Gorilla Glass 5
Protección IP68

También te recomendamos

Lavadora con función vapor: qué es y cómo debe utilizarse

El elogiado modo retrato del Pixel 2 llegará a todo tipo de Android... si los fabricantes quieren

El Google Pixel 3 XL se deja ver en Geekbench con 4GB de RAM y Android 9.0 P

-
La noticia Del Google Pixel 2 al Google Pixel 3: todo lo que ha cambiado fue publicada originalmente en Xataka Móvil por Samuel Fernández .


          A Look at CNTK v2.6 and the Iris Dataset      Cache   Translate Page      

Version 2.6 of CNTK was released a few weeks ago so I figured I’d update my system and give it a try. CNTK (“Cognitive Network Tool Kit”) is Microsoft’s neural network code library. Primary alternatives include Google’s TensorFlow and Keras (a library that makes TF easier to use), and Facebook’s PyTorch.

To cut to the chase, I deleted by existing CNTK and then installed v2.6 using the pip utility, and then . .

As I write this, I think back about all the effort that was required to figure out how to install CNTK (and TF and Keras and PyTorch). It’s easy for me now, but if you’re new to using neural network code libraries, trust me, there’s a lot to learn ― mostly about all the many things that can go wrong with an installation, how to interpret the error messages, and how to resolve.

OK, back to my post. I ran my favorite demo, classification on the Iris Dataset. My old (written for v2.5) CNTK code ran as expected. Excellent!


A Look at CNTK v2.6 and the Iris Dataset

The real moral of the story is that deep learning with neural network libraries is new and still in a state of constant flux. This makes it tremendously difficult to stay abreast of changes. New releases of these libraries emerge not every free months, or even every few weeks, but often every few days. The pace of development is unlike anything I’ve ever seen in computer science.


A Look at CNTK v2.6 and the Iris Dataset

Additionally, the NN libraries are just the tip of the technology pyramid. There are dozens and dozens of supporting systems, and they are being developed with blazing speed too. For example, I did an Internet search for “auto ML” and found many systems that are wrappers over CNTK or TF/Keras or PyTorch, and that are intended to automate the process pipeline of things like hyperparameter tuning, data preprocessing, and so on.

The blistering pace of development of neural network code libraries and supporting software will eventually slow down (maybe 18 months as a wild guess), but for now it’s an incredibly exciting time to be working with deep learning systems.


A Look at CNTK v2.6 and the Iris Dataset

I suspect that an artist’s style doesn’t change too quickly over time (well, after his/her formative years). Three paintings by an unknown (to me) artist with similar compositions but slightly different styles.


          Document worth reading: “Deep Learning for Generic Object Detection: A Survey”      Cache   Translate Page      
Generic object detection, aiming at locating object instances from a large number of predefined categories in natural images, is one …

Continue reading


          JR0081235 - Deep Learning R&D Engineer      Cache   Translate Page      
Intel - Bangalore, Karnataka - - Deep Learning R&D engineer to model the performance of different deep learning workloads on customized hardware. Work with design...
          Deep Learning Is Creating A Competitive Edge For Social Traders      Cache   Translate Page      

Description: Just like cryptocurrencies took the world by storm, social trading is making strides in becoming the hot trend in the world of trading. Is Social Trading the Next Big Thing? Deep learning algorithms have played a very important role in the evolution of social media. There are a number of other applications for these […]

The post Deep Learning Is Creating A Competitive Edge For Social Traders appeared first on SmartData Collective.


          Expertos en tecnologías clave plantean grupos de trabajo para compartir y acelerar el desarrollo de algoritmos basados en 'deep learning'      Cache   Translate Page      
Valencia, 10/10/2018 El Comité Estratégico de Innovación Especializado (CEIE) en Tecnologías Habilitadoras para la nueva economía ha propuesto la generación de grupos de trabajo donde las empresas...

[[ This is a content summary only. Visit my website for full links, other content, and more! ]]

          ESC Is Ready to Cast Its Spell on Minneapolis       Cache   Translate Page      

Take a spin in an autonomous vehicle, visit top companies at their exhibitor booths, learn the latest in embedded systems, IoT, and advanced tech from some of the industry’s leading experts…These are just three of several reasons to register for and attend ESC Minneapolis, which is less than one month away!

Click Here to Register Today!

Let’s start with the ESC Minneapolis keynotes. Michael McAlpine is a professor of mechanical engineering at the University of Minnesota, where he researches 3D printing functional materials. Join him in Engineering HQ on Wednesday, Oct. 31 for 3D Printing Functional Materials & Devices. On Thursday, Nov. 1, VSI Labs’ Phil Magney will be in the theater talking about The Future of Automated Driving. And that’s not all! VSI Labs will also feature its Level 2 autonomous vehicle in its booth on the Expo floor (#222) both days of the event. Stop by to check it out and to take a spin!

Beyond the keynotes, Engineering HQ will host dynamic panel discussions on artificial intelligence, augmented & virtual reality, and 3D printing and packaging. ABB Robotics VP, Dwight Morgan, will also host a discussion on the state of the robotics industry. And on Halloween, consultant Rob Reilly will demonstrate how he brought Hedley, his spooky robotic skull, to life with a smart machine sensor, Raspberry Pi, and some Arduinos, servos, and major creativity.

ESC’s comprehensive conference continues upstairs, with two full days of education in four tracks: Embedded Hardware, Embedded Software, IoT & Connected Devices, and Advanced Technologies.

Highlights from the Embedded Hardware track include:

            • Everything You Wanted to Know About ASICs, But Were Afraid to Ask

            • Selecting the Best Embedded Hardware for Your Product Design

            • How to Design Mission Critical FPGA Systems

            • Sensor Design with Bluetooth 5 Mesh Technology

Highlights from the Embedded Software track include:

            • An Introduction to RTOSs

            • ARM Cortex-M and RTOSs Are Meant for Each Other

            • Representing Memory-Mapped Hardware in C++

            • Write & Test Your C Code Sooner without Hardware

Highlights from the IoT & Connected Devices track include:

            • Wireless Sensor for Vehicle AdHoc Network Localization & Safety

            • System Design Trade-offs Beyond the Device

            • Talk to Your IoT Application: Android App Programming Demystified

            • Automating Power Management in MCU-Based IoT Nodes

Highlights from the Advanced Technologies track include:

            • How to Create an Embedded Vision System

            • The Functional Safety in Autonomous Vehicles Is Not an Afterthought

            • Object Detection Using LiDAR

            • Deep Learning Accelerators for Client Systems

 

Click Here for More Information on the Full ESC Minneapolis Line-up

And click here for a full list of exhibitors including: Green Hills Software, Rohde & Schwarz, Tektronix, Teledyne LeCroy, Siemens, and Proto Labs.

Finally, if you are planning to be at the show—and we hope you are—be sure to download the free event app to make the most of your show experience. Simply go to the App Store and search, “UBM Minneapolis.”

See you there!


          Comment on Deep Learning Models for Human Activity Recognition by Gábor Stikkel      Cache   Translate Page      
Sorry for that, here is the correct one: https://github.com/deadskull7/Human-Activity-Recognition-with-Neural-Network-using-Gyroscopic-and-Accelerometer-variables
          Comment on How to Train a Final Machine Learning Model by Xu Zhang      Cache   Translate Page      
Thank you so much for your great article. I understood that we should use all the data which we have to train our final model. However, when should I stop training when I train my final model with dataset including train+validation+test? Especially for a deep learning model. Let me explain it with an example: I have a CNN model with 100,000 examples. I will do the following procedure: 1. I split this dataset into training data 80,000, validation data 10,000 and test data 10,000. 2. I used my validation dataset to guide my training and hyperparameter tuning. Here I used early stopping to prevent overfitting. 3. Then I got my best performance and hyperparameters. From early stopping setting, I got that when I trained my model 37 epochs, the losses were low and performance using test data to evaluate was good. 4 I will finalize my model, train my final model with all my 100,000 data. Here is a problem. Without validation dataset, how can I know when I should stop training, that is how many epochs I should choose when I train my final models. Will I use the same epochs which are used before finalizing the model? or should I match the loss which I got before? I think for the machine learning models without early stopping training, they are no problems. But like deep learning models, when to stop training is a critical issue. Any advice? Thanks.
          Machine Vision and Learning for eXplainable AI (XAI)      Cache   Translate Page      
NASA Langley Research Center is seeking a postdoctoral researcher with ability in computer vision to develop learning methods (including deep learning) using electro-optical sensors to explore novel applications of autonomous systems, with an emphasis on small Unmanned Aerial Systems (sUAS). The successful applicant should hold a doctoral degree in a field related to computer vision and machi
          Step Size Matters in Deep Learning. (arXiv:1805.08890v2 [cs.LG] UPDATED)      Cache   Translate Page      

Authors: Kamil Nar, S. Shankar Sastry

Training a neural network with the gradient descent algorithm gives rise to a discrete-time nonlinear dynamical system. Consequently, behaviors that are typically observed in these systems emerge during training, such as convergence to an orbit but not to a fixed point or dependence of convergence on the initialization. Step size of the algorithm plays a critical role in these behaviors: it determines the subset of the local optima that the algorithm can converge to, and it specifies the magnitude of the oscillations if the algorithm converges to an orbit. To elucidate the effects of the step size on training of neural networks, we study the gradient descent algorithm as a discrete-time dynamical system, and by analyzing the Lyapunov stability of different solutions, we show the relationship between the step size of the algorithm and the solutions that can be obtained with this algorithm. The results provide an explanation for several phenomena observed in practice, including the deterioration in the training error with increased depth, the hardness of estimating linear mappings with large singular values, and the distinct performance of deep residual networks.


          Deep Learning-Experte/Expertin - Bosch Gruppe - Vaihingen an der Enz      Cache   Translate Page      
Standort Stuttgart-Vaihingen Arbeitsbereiche Informationstechnologie Einstieg als Berufserfahrene/-r Startdatum Nach Vereinbarung Arbeitszeit Vollzeit und...
Gefunden bei Bosch Gruppe - Fri, 07 Sep 2018 12:02:46 GMT - Zeige alle Vaihingen an der Enz Jobs
          Deep Learning Expert - Bosch Gruppe - Vaihingen an der Enz      Cache   Translate Page      
Standort Stuttgart-Vaihingen Arbeitsbereiche Informationstechnologie Einstieg als Berufserfahrene/-r Startdatum Nach Vereinbarung Arbeitszeit Vollzeit und...
Gefunden bei Bosch Gruppe - Fri, 07 Sep 2018 12:02:45 GMT - Zeige alle Vaihingen an der Enz Jobs
          Understanding AI: A Simple Explanation of AI, Machine Learning and Deep Learning      Cache   Translate Page      
Almost every piece of media content covering the latest technology in the world mentions AI at some point. Is there any piece of tech that includes coding not described as AI, Machine Learning or Deep Learning these days? If there is, it’s an exception to the rule. So when did software all become AI? What […]
          Deep Learning-Experte/Expertin - Bosch Gruppe - Vaihingen an der Enz      Cache   Translate Page      
Standort Stuttgart-Vaihingen Arbeitsbereiche Informationstechnologie Einstieg als Berufserfahrene/-r Startdatum Nach Vereinbarung Arbeitszeit Vollzeit und...
Gefunden bei Bosch Gruppe - Fri, 07 Sep 2018 12:02:46 GMT - Zeige alle Vaihingen an der Enz Jobs
          Deep Learning Expert - Bosch Gruppe - Vaihingen an der Enz      Cache   Translate Page      
Standort Stuttgart-Vaihingen Arbeitsbereiche Informationstechnologie Einstieg als Berufserfahrene/-r Startdatum Nach Vereinbarung Arbeitszeit Vollzeit und...
Gefunden bei Bosch Gruppe - Fri, 07 Sep 2018 12:02:45 GMT - Zeige alle Vaihingen an der Enz Jobs
          IBM and NVIDIA Collaborate to Expand Open Source Machine Learning Tools for Data Scientists      Cache   Translate Page      
With IBM's vast portfolio of deep learning and machine learning solutions, it is best positioned to bring this open-source technology to data scientists ...
          IBM and NVIDIA Collaborate to Expand Open Source Machine Learning Tools for Data Scientists      Cache   Translate Page      
With IBM's vast portfolio of deep learning and machine learning solutions, it is best positioned to bring this open-source technology to data scientists ...
          Probamos a fondo la tecnología DLSS Nvidia      Cache   Translate Page      

¿Qué ocurriría si los fabricantes de hardware para PC adoptasen las inteligentes tecnologías de reescalado que hoy en día son habituales en consola? Es un tema que hemos estudiado en el pasado, pero con el nuevo Deep Learning Super-Sampling (DLSS) de Nvidia ahora tenemos una tecnología de reconstrucción acelerada totalmente por hardware, produciendo unos resultados más que interesantes. De hecho, basándonos en la demo de Final Fantasy XV a la que hemos tenido acceso, DLSS incrementa el rendimiento un 40% y en cierto aspectos incluso mejora la calidad de la imagen.

¿Pero cómo funciona? En la presentación de la tecnología RTX durante la pasada Gamescom, el presidente de Nvidia explicó cómo la tecnología de deep learning - el centro de los nuevos núcleos tensores dentro de Turing - podían 'inferir' más detalle en cualquier imagen a través de la experiencia de aprendizaje previa al estudiar imágenes similares. En DLSS esto se traduce en que el supercomputador interno de Nvidia, llamado Saturn 5, analiza imágenes extremadamente detalladas de los juegos, produciendo un algoritmo con un tamaño de apenas unos megabytes que se descarga a través de una actualización de los drivers a la tarjeta RTX.

El juego se renderiza a una resolución menor y, al igual que en esas técnicas de mejora de imagen que funcionan tan bien gracias al deep learning, DLSS produce una imagen de mayor resolución. Pero estamos seguros de que aquí hay más cosas de las que dice Nvidia. Para empezar, DLSS depende de juegos que utilicen anti-aliasing temporal (lo cual, siendo justos, cubre prácticamente todos los motores actuales). Esto sugiere que DLSS extrae información de los anteriores frames para ayudar en su reconstrucción y en las mejoras que introduce con su algoritmo.

Read more…


          MIMO Matlab project      Cache   Translate Page      
communication MIMO machine learning deep learning Telecommunication fading channel (Budget: $30 - $250 USD, Jobs: Electrical Engineering, Electronics, Engineering, Matlab and Mathematica, Telecommunications Engineering)
          New Intel Vision Accelerator Solutions Speed Up Deep Learning and Artificial Intelligence on Edge Devices      Cache   Translate Page      
Today, Intel unveiled its family of Intel Vision Accelerator Design Products targeted at artificial intelligence (AI) inference and analytics performance on edge devices.
          Machine Vision and Learning for eXplainable AI (XAI)      Cache   Translate Page      
NASA Langley Research Center is seeking a postdoctoral researcher with ability in computer vision to develop learning methods (including deep learning) using electro-optical sensors to explore novel applications of autonomous systems, with an emphasis on small Unmanned Aerial Systems (sUAS). The successful applicant should hold a doctoral degree in a field related to computer vision and machi
          Bayou: l’AI che programma da sola      Cache   Translate Page      

Bayou è una deep learning software coding application sviluppata dal DARPA e da Google in grado di programmare da sola.

Leggi Bayou: l’AI che programma da sola


          Associate / Full Professor – Head, Division of Breast Imaging - Sunnybrook Health Sciences Centre - Toronto, ON      Cache   Translate Page      
Anne Martel, another SRI scientist has created a research program in machine/deep learning with applications in breast cancer. Breast Imager and Division Head....
From Indeed - Tue, 17 Jul 2018 17:17:53 GMT - View all Toronto, ON jobs
          Amazon desecha una IA de reclutamiento por su sesgo contra las mujeres      Cache   Translate Page      

Amazon desecha una IA de reclutamiento por su sesgo contra las mujeres#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

La inteligencia artificial está produciendo avances enormes en muchas ramas a un ritmo hasta ahora desconocido. Sin embargo, no se debe olvidar que bajo determinadas circunstancias y casos, debido a problemas generalmente relacionados con los datos usados en la base de datos de entrenamiento del software, su uso puede llegar a jugar malas pasadas. Amazon lo ha sufrido con una herramienta que emplea inteligencia artificial para reclutamiento laboral, según informa Reuters.

Estos sistemas ofrecen grandes ventajas para seleccionar a candidatos con mayor eficiencia, pero en el caso de la inteligencia artificial de Amazon, los resultados que ofrecía presentaban sesgos contra las mujeres, es decir, que por ejemplo, en reclutamiento de candidatos para trabajos de desarrollo de software y en otras áreas técnicas, el sistema no estaba aplicando principios de igualdad y meritocracia, sino que producía una importante discriminación contra las mujeres.

La inteligencia artificial presenta retos enormes incluso para gigantes como Amazon

El hecho de que una situación así se dé dentro de una empresa como Amazon, referente en el campo de la automatización y el uso de herramientas de aprendizaje automático, puede parecer inexplicable, pero no lo es. La razón detrás del sesgo contra el sexo femenino está en cómo ha funcionado el reclutamiento en Amazon en los últimos 10 años.

La inteligencia artificial debe ser alimentada de datos para generar modelos y encontrar similitudes positivas. Si el sistema de captación de empleados de la compañía en los últimos años generaba mayoritariamente contrataciones masculinas, y esos datos se le ofrecen a la herramienta sin indicaciones de lo que está bien o mal, o de la igualdad de oportunidades que se busca, el software interpreta que lo corriente es lo positivo. Acto seguido, actuará en base a ello.

En este caso, se crearon 500 modelos informáticos con atención a funciones específicas del empleo y localizaciones. Cada uno de esos 500 aprendió a reconocer 50.000 palabras que aparecían en el currículum de los candidatos. El problema es que los hombres, según la fuente de Reuters, tienden a usar términos como realización o conquista más que las mujeres, y el sistema los primó.

Si no se hace nada para frenarlo, la inteligencia artificial puede reproducir y amplificar los sesgos humanos

La simple introducción del término "women’s" en el currículum generaba un agravio contra las mujeres. Las mujeres tituladas en dos universidades femeninas no mixtas también sufrían efectos de degradación para el sistema. Aunque dentro de la empresa esto se vio como algo negativo y se quiso modificar ese comportamiento de la inteligencia artificial para hacerla más neutra, el proyecto se desechó cuando se percibió que no había certeza de que el sesgo no se pudiera reproducir en otro sentido.

El equipo de desarrolladores a cargo fue disuelto a causa de una pérdida de esperanza en el proyecto. La herramienta se llegó a usar, pero no para basarse únicamente en sus recomendaciones.

Un problema con difícil solución

Entrenar a un software de inteligencia artificial para obtener resultados óptimos no es algo banal. Como se ha mencionado, no es lo mismo alimentar a una máquina con datos a discreción, sin aportar más profundidad al análisis de datos, que hacerlo. Lo segundo es mucho más complicado, porque en este caso en Amazon se ha visto mal el sesgo, pero en otros muchos puede no ser así y que la desigualdad se perpetúe e incluso crezca a largo plazo.

Al diseñar estas herramientas las empresas no sólo deben buscar la productividad, sino de alguna forma reconocer qué patrones repetitivos y del todo positivos se pueden estar dando con decisiones por personas, para así generar modelos que tiendan más a la centralidad.

Puedes aprender más sobre inteligencia artificial escuchando nuestro podcast ‘Captcha’, donde debatimos en profundidad sobre todo lo que le rodea.

También te recomendamos

Lavadora con función vapor: qué es y cómo debe utilizarse

El último algoritmo de Google no predice lo que buscas sino cuando morirás

Crean una inteligencia artificial con la que afirman que se puede predecir tu personalidad escaneando tus ojos

-
La noticia Amazon desecha una IA de reclutamiento por su sesgo contra las mujeres fue publicada originalmente en Genbeta por Antonio Sabán .


          Deep Learning и задача понимания естественного языка      Cache   Translate Page      
Искусственный интеллект и Машинное обучение
Статистика : 132 Ответы || 2081 Просмотры Last post by Yodine
          Artificial Intelligence Gains Ground in Thyroid Nodule Diagnosis      Cache   Translate Page      
An AI system, programmed with 'deep learning' to evaluate thyroid nodule ultrasounds, shows impressive accuracy, particularly when combined with the judgment of junior physicians.
Medscape Medical News
          Workshop on Deep Learning for Recommender Systems [electronic resource]      Cache   Translate Page      
none
          Google AI Residency Program 2019: Research Training Role for Graduates in STEM Fields      Cache   Translate Page      
The Google AI Residency Program — previously known as the Google Brain Residency Program — is a 12-month research training role designed to jumpstart or advance your career in machine learning research. The goal of the residency is to help residents become productive and successful AI researchers.The Google AI Residency Program was created in 2015 with the goal of training and supporting the next generation of deep learning researchers. Residents will have the opportunity to be mentored by distinguished scientists and engineers from various teams within Google AI, and work on real-world machine learning problems and applications. In addition, they will also have the opportunity to collaborate and partner closely with various research groups across Google and Alphabet.As part of this program, Residents collaborate with distinguished scientists from various Google AI teams working on machine learning applications and problems. Residents have the opportunity to do everything from conducting fundamental research to contributing to products and services used by billions of people. We also encourage our Residents to publish their work externally. Take a look at the impactful research done by earlier cohorts.Residency LocationsBay Area (Mountain View and San Francisco)New York City, Cambridge (Massachusetts)Montreal and Toronto, CanadaSeattle (Washington State)Accra (Ghana)Tel Aviv (Israel)Zurich (Switzerland).Residents are placed based on interest, project fit, location preference and team needs. All are expected to work on site.The Google AI Residency Program will have 3 start dates over the course of 5 months, from June to October 2019. Exact dates are yet to be determined.

Apply at https://ngcareers.com/job/2018-10/google-ai-residency-program-2019-research-training-role-for-graduates-in-stem-fields-639/


          Neuton: A new, disruptive neural network framework for AI applications      Cache   Translate Page      
Deep learning neural networks are behind much of the progress in AI these days. Neuton is a new framework that claims to be much faster and more compact, and it requires less skills and training than anything the AWSs, Googles, and Facebooks of the world have.
          Artificial Intelligence Gains Ground in Thyroid Nodule Diagnosis Artificial Intelligence Gains Ground in Thyroid Nodule Diagnosis      Cache   Translate Page      
An AI system, programmed with'deep learning'to evaluate thyroid nodule ultrasounds, shows impressive accuracy, particularly when combined with the judgment of junior physicians.Medscape Medical News (Source: Medscape Medical News Headlines)
          Machine Vision and Learning for eXplainable AI (XAI)      Cache   Translate Page      
NASA Langley Research Center is seeking a postdoctoral researcher with ability in computer vision to develop learning methods (including deep learning) using electro-optical sensors to explore novel applications of autonomous systems, with an emphasis on small Unmanned Aerial Systems (sUAS). The successful applicant should hold a doctoral degree in a field related to computer vision and machi
          Advances in deep learning for drug discovery and biomarker development published in top journal      Cache   Translate Page      
(InSilico Medicine, Inc.) Insilico Medicine, one of the industry leaders bridging deep learning for biology, chemistry and digital medicine, announced the publication of a special issue dedicated to 'Deep Learing for Drug Discovery and Biomarker Development' in one of the top industry journals celebrating its 15th anniversary published by the American Chemical Society, Molecular Pharmaceutics. The special issue starts with an article by the founder and CEO of Insilico Medicine, Alex Zhavoronkov, PhD, titled "Artificial Intelligence for Drug Discovery, Biomarker Development, and Generation of Novel Chemistry".
           Jetson-equipped Teal One drone takes flight as a 60-mph Raspberry Pi in the sky       Cache   Translate Page      

It looks like a compact, fun drone much like any other, but the Teal One packs ...#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

It's taken a couple of years, but 20-year-old drone entrepreneur George Matus has finally got his flagship Teal One drone into production. Fast, compact and smart, the Teal One packs an Nvidia Jetson TX module and has its own app platform to enable AR, autonomy and deep learning applications.

.. Continue Reading Jetson-equipped Teal One drone takes flight as a 60-mph Raspberry Pi in the sky

Category: Drones

Tags:
          Amazon desecha una IA de reclutamiento por su sesgo contra las mujeres      Cache   Translate Page      

Amazon desecha una IA de reclutamiento por su sesgo contra las mujeres#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

La inteligencia artificial está produciendo avances enormes en muchas ramas a un ritmo hasta ahora desconocido. Sin embargo, no se debe olvidar que bajo determinadas circunstancias y casos, debido a problemas generalmente relacionados con los datos usados en la base de datos de entrenamiento del software, su uso puede llegar a jugar malas pasadas. Amazon lo ha sufrido con una herramienta que emplea inteligencia artificial para reclutamiento laboral, según informa Reuters.

Estos sistemas ofrecen grandes ventajas para seleccionar a candidatos con mayor eficiencia, pero en el caso de la inteligencia artificial de Amazon, los resultados que ofrecía presentaban sesgos contra las mujeres, es decir, que por ejemplo, en reclutamiento de candidatos para trabajos de desarrollo de software y en otras áreas técnicas, el sistema no estaba aplicando principios de igualdad y meritocracia, sino que producía una importante discriminación contra las mujeres.

La inteligencia artificial presenta retos enormes incluso para gigantes como Amazon

El hecho de que una situación así se dé dentro de una empresa como Amazon, referente en el campo de la automatización y el uso de herramientas de aprendizaje automático, puede parecer inexplicable, pero no lo es. La razón detrás del sesgo contra el sexo femenino está en cómo ha funcionado el reclutamiento en Amazon en los últimos 10 años.

La inteligencia artificial debe ser alimentada de datos para generar modelos y encontrar similitudes positivas. Si el sistema de captación de empleados de la compañía en los últimos años generaba mayoritariamente contrataciones masculinas, y esos datos se le ofrecen a la herramienta sin indicaciones de lo que está bien o mal, o de la igualdad de oportunidades que se busca, el software interpreta que lo corriente es lo positivo. Acto seguido, actuará en base a ello.

En este caso, se crearon 500 modelos informáticos con atención a funciones específicas del empleo y localizaciones. Cada uno de esos 500 aprendió a reconocer 50.000 palabras que aparecían en el currículum de los candidatos. El problema es que los hombres, según la fuente de Reuters, tienden a usar términos como realización o conquista más que las mujeres, y el sistema los primó.

Si no se hace nada para frenarlo, la inteligencia artificial puede reproducir y amplificar los sesgos humanos

La simple introducción del término "women’s" en el currículum generaba un agravio contra las mujeres. Las mujeres tituladas en dos universidades femeninas no mixtas también sufrían efectos de degradación para el sistema. Aunque dentro de la empresa esto se vio como algo negativo y se quiso modificar ese comportamiento de la inteligencia artificial para hacerla más neutra, el proyecto se desechó cuando se percibió que no había certeza de que el sesgo no se pudiera reproducir en otro sentido.

El equipo de desarrolladores a cargo fue disuelto a causa de una pérdida de esperanza en el proyecto. La herramienta se llegó a usar, pero no para basarse únicamente en sus recomendaciones.

Un problema con difícil solución

Entrenar a un software de inteligencia artificial para obtener resultados óptimos no es algo banal. Como se ha mencionado, no es lo mismo alimentar a una máquina con datos a discreción, sin aportar más profundidad al análisis de datos, que hacerlo. Lo segundo es mucho más complicado, porque en este caso en Amazon se ha visto mal el sesgo, pero en otros muchos puede no ser así y que la desigualdad se perpetúe e incluso crezca a largo plazo.

Al diseñar estas herramientas las empresas no sólo deben buscar la productividad, sino de alguna forma reconocer qué patrones repetitivos y del todo positivos se pueden estar dando con decisiones por personas, para así generar modelos que tiendan más a la centralidad.

Puedes aprender más sobre inteligencia artificial escuchando nuestro podcast ‘Captcha’, donde debatimos en profundidad sobre todo lo que le rodea.


          Machine Vision and Learning for eXplainable AI (XAI)      Cache   Translate Page      
NASA Langley Research Center is seeking a postdoctoral researcher with ability in computer vision to develop learning methods (including deep learning) using electro-optical sensors to explore novel applications of autonomous systems, with an emphasis on small Unmanned Aerial Systems (sUAS). The successful applicant should hold a doctoral degree in a field related to computer vision and machi
          Global AI in Telecommunication Market 2018 - 2025 : Trade Overview, Applications Analysis and Key Players - IBM. Microsoft, Intel, Google      Cache   Translate Page      

Qyresearchreports include new market research report "Global AI In Telecommunication Market Size, Status and Forecast 2025" to its huge collection of research reports.

Brooklyn, NY -- (SBWIRE) -- 10/10/2018 -- This report studies the global AI In Telecommunication market size, industry status and forecast, competition landscape and growth opportunity. This research report categorizes the global AI In Telecommunication market by companies, region, type and end-use industry.

The increasing adoption of AI for various applications in the telecommunication industry and the utilization of AI-enabled smartphones are expected to be driving the growth of the AI in telecommunication market. Incompatibility concerns between the AI technology and telecommunication systems, which may generate integration complexities in AI in telecommunication solutions, are expected to act as restraints for the growth of the market.
In 2017, the global AI In Telecommunication market size was xx million US$ and it is expected to reach xx million US$ by the end of 2025, with a CAGR of xx% during 2018-2025.

Get Free Report Sample and Customization: https://www.qyresearchreports.com/sample/sample.php?rep_id=1856812&type=S

This report focuses on the global top players, covered
IBM
Microsoft
Intel
Google
AT&T
Cisco Systems
Nuance Communications
Sentient Technologies
H2O.ai
Infosys
Salesforce
Nvidia
Others

Market segment by Regions/Countries, this report covers
United States
Europe
China
Japan
Southeast Asia
India

View Table Of Content : https://www.qyresearchreports.com/report/global-ai-in-telecommunication-market-size-status-and-forecast-2025.htm/toc

Market segment by Type, the product can be split into
Machine Learning and Deep Learning
Natural Language Processing

Market segment by Application, split into
Customer Analytics
Network Security
Network Optimization
Self-Diagnostics
Virtual Assistance
Others

The study objectives of this report are:
To study and forecast the market size of AI In Telecommunication in global market.
To analyze the global key players, SWOT analysis, value and global market share for top players.
To define, describe and forecast the market by type, end use and region.
To analyze and compare the market status and forecast between China and major regions, namely, United States, Europe, China, Japan, Southeast Asia, India and Rest of World.
To analyze the global key regions market potential and advantage, opportunity and challenge, restraints and risks.
To identify significant trends and factors driving or inhibiting the market growth.
To analyze the opportunities in the market for stakeholders by identifying the high growth segments.
To strategically analyze each submarket with respect to individual growth trend and their contribution to the market
To analyze competitive developments such as expansions, agreements, new product launches, and acquisitions in the market
To strategically profile the key players and comprehensively analyze their growth strategies.

In this study, the years considered to estimate the market size of AI In Telecommunication are as follows:
History Year: 2013-2017
Base Year: 2017
Estimated Year: 2018
Forecast Year 2018 to 2025
For the data information by region, company, type and application, 2017 is considered as the base year. Whenever data information was unavailable for the base year, the prior year has been considered.

Read Report Desrciption And TOC : https://www.qyresearchreports.com/report/global-ai-in-telecommunication-market-size-status-and-forecast-2025.htm

Key Stakeholders
AI In Telecommunication Manufacturers
AI In Telecommunication Distributors/Traders/Wholesalers
AI In Telecommunication Subcomponent Manufacturers
Industry Association
Downstream Vendors

Available Customizations
With the given market data, QYResearch offers customizations according to the company's specific needs. The following customization options are available for the report:
Regional and country-level analysis of the AI In Telecommunication market, by end-use.
Detailed analysis and profiles of additional market players.

About QYResearchReports.com
QYResearchReports.com delivers the latest strategic market intelligence to build a successful business footprint in China. Our syndicated and customized research reports provide companies with vital background information of the market and in-depth analysis on the Chinese trade and investment framework, which directly affects their business operations.

Contact US:
Brooklyn, NY 11230
United States
Toll Free: 866-997-4948 (USA-CANADA)
Tel: +1-518-618-1030
Email: sales@qyresearchreports.com

For more information on this press release visit: http://www.sbwire.com/press-releases/global-ai-in-telecommunication-market-2018-2025-trade-overview-applications-analysis-and-key-players-ibm-microsoft-intel-google-1061588.htm

Media Relations Contact

Ivan Gary
Manager
QYResearchreports
Telephone: 1-866-997-4948
Email: Click to Email Ivan Gary
Web: https://www.qyresearchreports.com/report/global-ai-in-telecommunication-market-size-status-and-forecast-2025.htm

#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000


          Ideanomics Closes on Purchase of University of Connecticut's Former West Hartford Campus      Cache   Translate Page      
WEST HARTFORD, Conn., Oct. 10, 2018 /PRNewswire/ -- Ideanomics (formerly: Seven Stars Cloud Group, Inc.) (NASDAQ: SSC) ("Ideanomics" or the "Company"), a leading global fintech and asset digitization services company which is establishing its global innovation center in West Hartford, is pleased to announce it has closed on its purchase of the 58-acre former University of Connecticut campus in West Hartford from the State of Connecticut. Ideanomics plans to transform the property into a world-renowned technology campus named Fintech Village. The planned $283 million-plus investment will focus on being an ultra high-speed computing facility and laboratory for developing new and leading edge Fintech solutions utilizing artificial intelligence, deep learning, IoT, and blockchain.
          «Deep learning»: L'homme prend sa première grosse raclée par la machine en matière de traduction      Cache   Translate Page      
Une intelligence artificielle a traduit « Deep Learning », considéré comme un ouvrage de référence du genre…
           Jetson-equipped Teal One drone takes flight as a 60-mph Raspberry Pi in the sky       Cache   Translate Page      

It looks like a compact, fun drone much like any other, but the Teal One packs ...#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

It's taken a couple of years, but 20-year-old drone entrepreneur George Matus has finally got his flagship Teal One drone into production. Fast, compact and smart, the Teal One packs an Nvidia Jetson TX module and has its own app platform to enable AR, autonomy and deep learning applications.

.. Continue Reading Jetson-equipped Teal One drone takes flight as a 60-mph Raspberry Pi in the sky

Category: Drones

Tags:
          Open Source RAPIDS GPU Platform to Accelerate Predictive Data Analytics      Cache   Translate Page      

Today NVIDIA announced a GPU-acceleration platform for data science and machine learning, with broad adoption from industry leaders, that enables even the largest companies to analyze massive amounts of data and make accurate business predictions at unprecedented speed. "It integrates seamlessly into the world’s most popular data science libraries and workflows to speed up machine learning. We are turbocharging machine learning like we have done with deep learning,” he said.

The post Open Source RAPIDS GPU Platform to Accelerate Predictive Data Analytics appeared first on insideHPC.


          Machine Learning / Algorithim Developer - TECHNICA CORPORATION - Dulles, VA      Cache   Translate Page      
Job Description: We are seeking a highly creative software engineer experienced in artificial intelligence and deep learning techniques to design, develop,...
From Technica Corporation - Fri, 05 Oct 2018 10:31:19 GMT - View all Dulles, VA jobs
          Deep Learning и задача понимания естественного языка      Cache   Translate Page      
Искусственный интеллект и Машинное обучение
Статистика : 135 Ответы || 2113 Просмотры Last post by Alephegg
          Comment on How to Setup a Python Environment for Machine Learning and Deep Learning with Anaconda by Sindhu      Cache   Translate Page      
theano: 1.0.3 tensorflow: 1.11.0 Using TensorFlow backend. keras: 2.2.4
          Comment on How to Setup a Python Environment for Machine Learning and Deep Learning with Anaconda by Sindhu      Cache   Translate Page      
scipy: 1.1.0 numpy: 1.15.2 matplotlib: 3.0.0 pandas: 0.23.4 statsmodels: 0.9.0 sklearn: 0.20.0
          NVIDIA HGX-2 Ai Supercomputer Comes to Oracle Cloud Infrastructure      Cache   Translate Page      

Today Oracle announced support for the NVIDIA HGX-2 platform on Oracle Cloud Infrastructure. Designed to meet the needs of the next generation of analytics, machine learning, and AI, Oracle is offering GPU-accelerated deep learning and HPC containers from the NVIDIA GPU Cloud container registry. “Whether you are an engineer, data scientist, researcher or developer, we are bringing the power of compute and cloud to your fingertips.

The post NVIDIA HGX-2 Ai Supercomputer Comes to Oracle Cloud Infrastructure appeared first on insideHPC.


          Exploring LSTMs      Cache   Translate Page      

It turns out LSTMs are a fairly simple extension to neural networks, and they're behind a lot of the amazing achievements deep learning has made in the past few years. So I'll try to present them as intuitively as possible – in such a way that you could have discovered them yourself.

But first, a picture:

LSTM

Aren't LSTMs beautiful? Let's go.

(Note: if you're already familiar with neural networks and LSTMs, skip to the middle – the first half of this post is a tutorial.)

Neural Networks

Imagine we have a sequence of images from a movie, and we want to label each image with an activity (is this a fight?, are the characters talking?, are the characters eating?).

How do we do this?

One way is to ignore the sequential nature of the images, and build a per-image classifier that considers each image in isolation. For example, given enough images and labels:

  • Our algorithm might first learn to detect low-level patterns like shapes and edges.
  • With more data, it might learn to combine these patterns into more complex ones, like faces (two circular things atop a triangular thing atop an oval thing) or cats.
  • And with even more data, it might learn to map these higher-level patterns into activities themselves (scenes with mouths, steaks, and forks are probably about eating).

This, then, is a deep neural network: it takes an image input, returns an activity output, and – just as we might learn to detect patterns in puppy behavior without knowing anything about dogs (after seeing enough corgis, we discover common characteristics like fluffy butts and drumstick legs; next, we learn advanced features like splooting) – in between it learns to represent images through hidden layers of representations.

Mathematically

I assume people are familiar with basic neural networks already, but let's quickly review them.

  • A neural network with a single hidden layer takes as input a vector x, which we can think of as a set of neurons.
  • Each input neuron is connected to a hidden layer of neurons via a set of learned weights.
  • The jth hidden neuron outputs \(h_j = \phi(\sum_i w_{ij} x_i)\), where \(\phi\) is an activation function.
  • The hidden layer is fully connected to an output layer, and the jth output neuron outputs \(y_j = \sum_i v_{ij} h_i\). If we need probabilities, we can transform the output layer via a softmax function.

In matrix notation:

$$h = \phi(Wx)$$
$$y = Vh$$

where

  • x is our input vector
  • W is a weight matrix connecting the input and hidden layers
  • V is a weight matrix connecting the hidden and output layers
  • Common activation functions for \(\phi\) are the sigmoid function, \(\sigma(x)\), which squashes numbers into the range (0, 1); the hyperbolic tangent, \(tanh(x)\), which squashes numbers into the range (-1, 1), and the rectified linear unit, \(ReLU(x) = max(0, x)\).

Here's a pictorial view:

Neural Network

(Note: to make the notation a little cleaner, I assume x and h each contain an extra bias neuron fixed at 1 for learning bias weights.)

Remembering Information with RNNs

Ignoring the sequential aspect of the movie images is pretty ML 101, though. If we see a scene of a beach, we should boost beach activities in future frames: an image of someone in the water should probably be labeled swimming, not bathing, and an image of someone lying with their eyes closed is probably suntanning. If we remember that Bob just arrived at a supermarket, then even without any distinctive supermarket features, an image of Bob holding a slab of bacon should probably be categorized as shopping instead of cooking.

So what we'd like is to let our model track the state of the world:

  1. After seeing each image, the model outputs a label and also updates the knowledge it's been learning. For example, the model might learn to automatically discover and track information like location (are scenes currently in a house or beach?), time of day (if a scene contains an image of the moon, the model should remember that it's nighttime), and within-movie progress (is this image the first frame or the 100th?). Importantly, just as a neural network automatically discovers hidden patterns like edges, shapes, and faces without being fed them, our model should automatically discover useful information by itself.
  2. When given a new image, the model should incorporate the knowledge it's gathered to do a better job.

This, then, is a recurrent neural network. Instead of simply taking an image and returning an activity, an RNN also maintains internal memories about the world (weights assigned to different pieces of information) to help perform its classifications.

Mathematically

So let's add the notion of internal knowledge to our equations, which we can think of as pieces of information that the network maintains over time.

But this is easy: we know that the hidden layers of neural networks already encode useful information about their inputs, so why not use these layers as the memory passed from one time step to the next? This gives us our RNN equations:

$$h_t = \phi(Wx_t + Uh_{t-1})$$
$$y_t = Vh_t$$

Note that the hidden state computed at time \(t\) (\(h_t\), our internal knowledge) is fed back at the next time step. (Also, I'll use concepts like hidden state, knowledge, memories, and beliefs to describe \(h_t\) interchangeably.)

RNN

Longer Memories through LSTMs

Let's think about how our model updates its knowledge of the world. So far, we've placed no constraints on this update, so its knowledge can change pretty chaotically: at one frame it thinks the characters are in the US, at the next frame it sees the characters eating sushi and thinks they're in Japan, and at the next frame it sees polar bears and thinks they're on Hydra Island. Or perhaps it has a wealth of information to suggest that Alice is an investment analyst, but decides she's a professional assassin after seeing her cook.

This chaos means information quickly transforms and vanishes, and it's difficult for the model to keep a long-term memory. So what we'd like is for the network to learn how to update its beliefs (scenes without Bob shouldn't change Bob-related information, scenes with Alice should focus on gathering details about her), in a way that its knowledge of the world evolves more gently.

This is how we do it.

  1. Adding a forgetting mechanism. If a scene ends, for example, the model should forget the current scene location, the time of day, and reset any scene-specific information; however, if a character dies in the scene, it should continue remembering that he's no longer alive. Thus, we want the model to learn a separate forgetting/remembering mechanism: when new inputs come in, it needs to know which beliefs to keep or throw away.
  2. Adding a saving mechanism. When the model sees a new image, it needs to learn whether any information about the image is worth using and saving. Maybe your mom sent you an article about the Kardashians, but who cares?
  3. So when new a input comes in, the model first forgets any long-term information it decides it no longer needs. Then it learns which parts of the new input are worth using, and saves them into its long-term memory.
  4. Focusing long-term memory into working memory. Finally, the model needs to learn which parts of its long-term memory are immediately useful. For example, Bob's age may be a useful piece of information to keep in the long term (children are more likely to be crawling, adults are more likely to be working), but is probably irrelevant if he's not in the current scene. So instead of using the full long-term memory all the time, it learns which parts to focus on instead.

This, then, is an long short-term memory network. Whereas an RNN can overwrite its memory at each time step in a fairly uncontrolled fashion, an LSTM transforms its memory in a very precise way: by using specific learning mechanisms for which pieces of information to remember, which to update, and which to pay attention to. This helps it keep track of information over longer periods of time.

Mathematically

Let's describe the LSTM additions mathematically.

At time \(t\), we receive a new input \(x_t\). We also have our long-term and working memories passed on from the previous time step, \(ltm_{t-1}\) and \(wm_{t-1}\) (both n-length vectors), which we want to update.

We'll start with our long-term memory. First, we need to know which pieces of long-term memory to continue remembering and which to discard, so we want to use the new input and our working memory to learn a remember gate of n numbers between 0 and 1, each of which determines how much of a long-term memory element to keep. (A 1 means to keep it, a 0 means to forget it entirely.)

Naturally, we can use a small neural network to learn this remember gate:

$$remember_t = \sigma(W_r x_t + U_r wm_{t-1}) $$

(Notice the similarity to our previous network equations; this is just a shallow neural network. Also, we use a sigmoid activation because we need numbers between 0 and 1.)

Next, we need to compute the information we can learn from \(x_t\), i.e., a candidate addition to our long-term memory:

$$ ltm'_t = \phi(W_l x_t + U_l wm_{t-1}) $$

\(\phi\) is an activation function, commonly chosen to be \(tanh\).

Before we add the candidate into our memory, though, we want to learn which parts of it are actually worth using and saving:

$$save_t = \sigma(W_s x_t + U_s wm_{t-1})$$

(Think of what happens when you read something on the web. While a news article might contain information about Hillary, you should ignore it if the source is Breitbart.)

Let's now combine all these steps. After forgetting memories we don't think we'll ever need again and saving useful pieces of incoming information, we have our updated long-term memory:

$$ltm_t = remember_t \circ ltm_{t-1} + save_t \circ ltm'_t$$

where \(\circ\) denotes element-wise multiplication.

Next, let's update our working memory. We want to learn how to focus our long-term memory into information that will be immediately useful. (Put differently, we want to learn what to move from an external hard drive onto our working laptop.) So we learn a focus/attention vector:

$$focus_t = \sigma(W_f x_t + U_f wm_{t-1})$$

Our working memory is then

$$wm_t = focus_t \circ \phi(ltm_t)$$

In other words, we pay full attention to elements where the focus is 1, and ignore elements where the focus is 0.

And we're done! Hopefully this made it into your long-term memory as well.


To summarize, whereas a vanilla RNN uses one equation to update its hidden state/memory:

$$h_t = \phi(Wx_t + Uh_{t-1})$$

An LSTM uses several:

$$ltm_t = remember_t \circ ltm_{t-1} + save_t \circ ltm'_t$$
$$wm_t = focus_t \circ tanh(ltm_t)$$

where each memory/attention sub-mechanism is just a mini brain of its own:

$$remember_t = \sigma(W_r x_t+ U_r wm_{t-1}) $$
$$save_t = \sigma(W_s x_t + U_s wm_{t-1})$$
$$focus_t = \sigma(W_f x_t + U_f wm_{t-1})$$
$$ ltm'_t = tanh(W_l x_t + U_l wm_{t-1}) $$

(Note: the terminology and variable names I've been using are different from the usual literature. Here are the standard names, which I'll use interchangeably from now on:

  • The long-term memory, \(ltm_t\), is usually called the cell state, denoted \(c_t\).
  • The working memory, \(wm_t\), is usually called the hidden state, denoted \(h_t\). This is analogous to the hidden state in vanilla RNNs.
  • The remember vector, \(remember_t\), is usually called the forget gate (despite the fact that a 1 in the forget gate still means to keep the memory and a 0 still means to forget it), denoted \(f_t\).
  • The save vector, \(save_t\), is usually called the input gate (as it determines how much of the input to let into the cell state), denoted \(i_t\).
  • The focus vector, \(focus_t\), is usually called the output gate, denoted \(o_t\). )

LSTM

Snorlax

I could have caught a hundred Pidgeys in the time it took me to write this post, so here's a cartoon.

Neural Networks

Neural Network

Recurrent Neural Networks

RNN

LSTMs

LSTM

Learning to Code

Let's look at a few examples of what an LSTM can do. Following Andrej Karpathy's terrific post, I'll use character-level LSTM models that are fed sequences of characters and trained to predict the next character in the sequence.

While this may seem a bit toyish, character-level models can actually be very useful, even on top of word models. For example:

  • Imagine a code autocompleter smart enough to allow you to program on your phone. An LSTM could (in theory) track the return type of the method you're currently in, and better suggest which variable to return; it could also know without compiling whether you've made a bug by returning the wrong type.
  • NLP applications like machine translation often have trouble dealing with rare terms. How do you translate a word you've never seen before, or convert adjectives to adverbs? Even if you know what a tweet means, how do you generate a new hashtag to capture it? Character models can daydream new terms, so this is another area with interesting applications.

So to start, I spun up an EC2 p2.xlarge spot instance, and trained a 3-layer LSTM on the Apache Commons Lang codebase. Here's a program it generates after a few hours.

While the code certainly isn't perfect, it's better than a lot of data scientists I know. And we can see that the LSTM has learned a lot of interesting (and correct!) coding behavior:

  • It knows how to structure classes: a license up top, followed by packages and imports, followed by comments and a class definition, followed by variables and methods. Similarly, it knows how to create methods: comments follow the correct orders (description, then @param, then @return, etc.), decorators are properly placed, and non-void methods end with appropriate return statements. Crucially, this behavior spans long ranges of code – see how giant the blocks are!
  • It can also track subroutines and nesting levels: indentation is always correct, and if statements and for loops are always closed out.
  • It even knows how to create tests.

How does the model do this? Let's look at a few of the hidden states.

Here's a neuron that seems to track the code's outer level of indentation:

(As the LSTM moves through the sequence, its neurons fire at varying intensities. The picture represents one particular neuron, where each row is a sequence and characters are color-coded according to the neuron's intensity; dark blue shades indicate large, positive activations, and dark red shades indicate very negative activations.)

Outer Level of Indentation

And here's a neuron that counts down the spaces between tabs:

Tab Spaces

For kicks, here's the output of a different 3-layer LSTM trained on TensorFlow's codebase:

There are plenty of other fun examples floating around the web, so check them out if you want to see more.

Investigating LSTM Internals

Let's dig a little deeper. We looked in the last section at examples of hidden states, but I wanted to play with LSTM cell states and their other memory mechanisms too. Do they fire when we expect, or are there surprising patterns?

Counting

To investigate, let's start by teaching an LSTM to count. (Remember how the Java and Python LSTMs were able to generate proper indentation!) So I generated sequences of the form

aaaaaXbbbbb

(N "a" characters, followed by a delimiter X, followed by N "b" characters, where 1 <= N <= 10), and trained a single-layer LSTM with 10 hidden neurons.

As expected, the LSTM learns perfectly within its training range – and can even generalize a few steps beyond it. (Although it starts to fail once we try to get it to count to 19.)

aaaaaaaaaaaaaaaXbbbbbbbbbbbbbbb
aaaaaaaaaaaaaaaaXbbbbbbbbbbbbbbbb
aaaaaaaaaaaaaaaaaXbbbbbbbbbbbbbbbbb
aaaaaaaaaaaaaaaaaaXbbbbbbbbbbbbbbbbbb
aaaaaaaaaaaaaaaaaaaXbbbbbbbbbbbbbbbbbb # Here it begins to fail: the model is given 19 "a"s, but outputs only 18 "b"s.

We expect to find a hidden state neuron that counts the number of a's if we look at its internals. And we do:

Neuron #2 Hidden State

I built a small web app to play around with LSTMs, and Neuron #2 seems to be counting both the number of a's it's seen, as well as the number of b's. (Remember that cells are shaded according to the neuron's activation, from dark red [-1] to dark blue [+1].)

What about the cell state? It behaves similarly:

Neuron #2 Cell State

One interesting thing is that the working memory looks like a "sharpened" version of the long-term memory. Does this hold true in general?

It does. (This is exactly as we would expect, since the long-term memory gets squashed by the tanh activation function and the output gate limits what gets passed on.) For example, here is an overview of all 10 cell state nodes at once. We see plenty of light-colored cells, representing values close to 0.

Counting LSTM Cell States

In contrast, the 10 working memory neurons look much more focused. Neurons 1, 3, 5, and 7 are even zeroed out entirely over the first half of the sequence.

Counting LSTM Hidden States

Let's go back to Neuron #2. Here are the candidate memory and input gate. They're relatively constant over each half of the sequence – as if the neuron is calculating a += 1 or b += 1 at each step.

Counting LSTM Candidate Memory

Input Gate

Finally, here's an overview of all of Neuron 2's internals:

Neuron 2 Overview

If you want to investigate the different counting neurons yourself, you can play around with the visualizer here.

(Note: this is far from the only way an LSTM can learn to count, and I'm anthropomorphizing quite a bit here. But I think viewing the network's behavior is interesting and can help build better models – after all, many of the ideas in neural networks come from analogies to the human brain, and if we see unexpected behavior, we may be able to design more efficient learning mechanisms.)

Count von Count

Let's look at a slightly more complicated counter. This time, I generated sequences of the form

aaXaXaaYbbbbb

(N a's with X's randomly sprinkled in, followed by a delimiter Y, followed by N b's). The LSTM still has to count the number of a's, but this time needs to ignore the X's as well.

Here's the full LSTM. We expect to see a counting neuron, but one where the input gate is zero whenever it sees an X. And we do!

Counter 2 - Cell State

Above is the cell state of Neuron 20. It increases until it hits the delimiter Y, and then decreases to the end of the sequence – just like it's calculating a num_bs_left_to_print variable that increments on a's and decrements on b's.

If we look at its input gate, it is indeed ignoring the X's:

Counter 2 - Input Gate

Interestingly, though, the candidate memory fully activates on the irrelevant X's – which shows why the input gate is needed. (Although, if the input gate weren't part of the architecture, presumably the network would have presumably learned to ignore the X's some other way, at least for this simple example.)

Counter 2 - Candidate Memory

Let's also look at Neuron 10.

Counter 2 - Neuron 10

This neuron is interesting as it only activates when reading the delimiter "Y" – and yet it still manages to encode the number of a's seen so far in the sequence. (It may be hard to tell from the picture, but when reading Y's belonging to sequences with the same number of a's, all the cell states have values either identical or within 0.1% of each other. You can see that Y's with fewer a's are lighter than those with more.) Perhaps some other neuron sees Neuron 10 slacking and helps a buddy out.

Remembering State

Next, I wanted to look at how LSTMs remember state. I generated sequences of the form

AxxxxxxYa
BxxxxxxYb

(i.e., an "A" or B", followed by 1-10 x's, then a delimiter "Y", ending with a lowercase version of the initial character). This way the network needs to remember whether it's in an "A" or "B" state.

We expect to find a neuron that fires when remembering that the sequence started with an "A", and another neuron that fires when remembering that it started with a "B". We do.

For example, here is an "A" neuron that activates when it reads an "A", and remembers until it needs to generate the final character. Notice that the input gate ignores all the "x" characters in between.

A Neuron - #8

Here is its "B" counterpart:

B Neuron - #17

One interesting point is that even though knowledge of the A vs. B state isn't needed until the network reads the "Y" delimiter, the hidden state fires throughout all the intermediate inputs anyways. This seems a bit "inefficient", but perhaps it's because the neurons are doing a bit of double-duty in counting the number of x's as well.

Copy Task

Finally, let's look at how an LSTM learns to copy information. (Recall that our Java LSTM was able to memorize and copy an Apache license.)

(Note: if you think about how LSTMs work, remembering lots of individual, detailed pieces of information isn't something they're very good at. For example, you may have noticed that one major flaw of the LSTM-generated code was that it often made use of undefined variables – the LSTMs couldn't remember which variables were in scope. This isn't surprising, since it's hard to use single cells to efficiently encode multi-valued information like characters, and LSTMs don't have a natural mechanism to chain adjacent memories to form words. Memory networks and neural Turing machines are two extensions to neural networks that help fix this, by augmenting with external memory components. So while copying isn't something LSTMs do very efficiently, it's fun to see how they try anyways.)

For this copy task, I trained a tiny 2-layer LSTM on sequences of the form

baaXbaa
abcXabc

(i.e., a 3-character subsequence composed of a's, b's, and c's, followed by a delimiter "X", followed by the same subsequence).

I wasn't sure what "copy neurons" would look like, so in order to find neurons that were memorizing parts of the initial subsequence, I looked at their hidden states when reading the delimiter X. Since the network needs to encode the initial subsequence, its states should exhibit different patterns depending on what they're learning.

The graph below, for example, plots Neuron 5's hidden state when reading the "X" delimiter. The neuron is clearly able to distinguish sequences beginning with a "c" from those that don't.

Neuron 5

For another example, here is Neuron 20's hidden state when reading the "X". It looks like it picks out sequences beginning with a "b".

Neuron 20 Hidden State

Interestingly, if we look at Neuron 20's cell state, it almost seems to capture the entire 3-character subsequence by itself (no small feat given its one-dimensionality!):

Neuron 20 Cell State

Here are Neuron 20's cell and hidden states, across the entire sequence. Notice that its hidden state is turned off over the entire initial subsequence (perhaps expected, since its memory only needs to be passively kept at that point).

Copy LSTM - Neuron 20 Hidden and Cell

However, if we look more closely, the neuron actually seems to be firing whenever the next character is a "b". So rather than being a "the sequence started with a b" neuron, it appears to be a "the next character is a b" neuron.

As far as I can tell, this pattern holds across the network – all the neurons seem to be predicting the next character, rather than memorizing characters at specific positions. For example, Neuron 5 seems to be a "next character is a c" predictor.

Copy LSTM - Neuron 5

I'm not sure if this is the default kind of behavior LSTMs learn when copying information, or what other copying mechanisms are available as well.

States and Gates

To really hone in and understand the purpose of the different states and gates in an LSTM, let's repeat the previous section with a small pivot.

Cell State and Hidden State (Memories)

We originally described the cell state as a long-term memory, and the hidden state as a way to pull out and focus these memories when needed.

So when a memory is currently irrelevant, we expect the hidden state to turn off – and that's exactly what happens for this sequence copying neuron.

Copy Machine

Forget Gate

The forget gate discards information from the cell state (0 means to completely forget, 1 means to completely remember), so we expect it to fully activate when it needs to remember something exactly, and to turn off when information is never going to be needed again.

That's what we see with this "A" memorizing neuron: the forget gate fires hard to remember that it's in an "A" state while it passes through the x's, and turns off once it's ready to generate the final "a".

Forget Gate

Input Gate (Save Gate)

We described the job of the input gate (what I originally called the save gate) as deciding whether or not to save information from a new input. Thus, it should turn off at useless information.

And that's what this selective counting neuron does: it counts the a's and b's, but ignores the irrelevant x's.

Input Gate

What's amazing is that nowhere in our LSTM equations did we specify that this is how the input (save), forget (remember), and output (focus) gates should work. The network just learned what's best.

Extensions

Now let's recap how you could have discovered LSTMs by yourself.

First, many of the problems we'd like to solve are sequential or temporal of some sort, so we should incorporate past learnings into our models. But we already know that the hidden layers of neural networks encode useful information, so why not use these hidden layers as the memories we pass from one time step to the next? And so we get RNNs.

But we know from our own behavior that we don't keep track of knowledge willy-nilly; when we read a new article about politics, we don't immediately believe whatever it tells us and incorporate it into our beliefs of the world. We selectively decide what information to save, what information to discard, and what pieces of information to use to make decisions the next time we read the news. Thus, we want to learn how to gather, update, and apply information – and why not learn these things through their own mini neural networks? And so we get LSTMs.

And now that we've gone through this process, we can come up with our own modifications.

  • For example, maybe you think it's silly for LSTMs to distinguish between long-term and working memories – why not have one? Or maybe you find separate remember gates and save gates kind of redundant – anything we forget should be replaced by new information, and vice-versa. And now you've come up with one popular LSTM variant, the GRU.
  • Or maybe you think that when deciding what information to remember, save, and focus on, we shouldn't rely on our working memory alone – why not use our long-term memory as well? And now you've discovered Peephole LSTMs.

Making Neural Nets Great Again

Let's look at one final example, using a 2-layer LSTM trained on Trump's tweets. Despite the tiny big dataset, it's enough to learn a lot of patterns.

For example, here's a neuron that tracks its position within hashtags, URLs, and @mentions:

Hashtags, URLs, @mentions

Here's a proper noun detector (note that it's not simply firing at capitalized words):

Proper Nouns

Here's an auxiliary verb + "to be" detector ("will be", "I've always been", "has never been"):

Modal Verbs

Here's a quote attributor:

Quotes

There's even a MAGA and capitalization neuron:

MAGA

And here are some of the proclamations the LSTM generates (okay, one of these is a real tweet):

Tweets Tweet

Unfortunately, the LSTM merely learned to ramble like a madman.

Recap

That's it. To summarize, here's what you've learned:

Candidate Memory

Here's what you should save:

Save

And now it's time for that donut.

Thanks to Chen Liang for some of the TensorFlow code I used, Ben Hamner and Kaggle for the Trump dataset, and, of course, Schmidhuber and Hochreiter for their original paper. If you want to explore the LSTMs yourself, feel free to play around!


          Associate / Full Professor – Head, Division of Breast Imaging - Sunnybrook Health Sciences Centre - Toronto, ON      Cache   Translate Page      
Anne Martel, another SRI scientist has created a research program in machine/deep learning with applications in breast cancer. Breast Imager and Division Head....
From Indeed - Tue, 17 Jul 2018 17:17:53 GMT - View all Toronto, ON jobs
           Jetson-equipped Teal One drone takes flight as a 60-mph Raspberry Pi in the sky       Cache   Translate Page      

It looks like a compact, fun drone much like any other, but the Teal One packs ...#source%3Dgooglier%2Ecom#https%3A%2F%2Fgooglier%2Ecom%2Fpage%2F%2F10000

It's taken a couple of years, but 20-year-old drone entrepreneur George Matus has finally got his flagship Teal One drone into production. Fast, compact and smart, the Teal One packs an Nvidia Jetson TX module and has its own app platform to enable AR, autonomy and deep learning applications.

.. Continue Reading Jetson-equipped Teal One drone takes flight as a 60-mph Raspberry Pi in the sky

Category: Drones

Tags:
          Streamr Partner With Daisy AI to Enhance Artificial Intelligence Machine Learning      Cache   Translate Page      
Crypto analytics firm CRYPTICS also becomes a vendor of Streamr Marketplace data Zug, Switzerland — October 10, 2018 — Streamr, the award-winning open-source, real-time data platform, has partnered with Japan-based Daisy AI, the AI platform using blockchain for deep learning. Daisy AI announces today that it will use Streamr as its official data provider to exclusively purchase data from the Streamr Marketplace. Recognizing the benefits of a decentralized marketplace model for transparent and secure information, Daisy AI is purchasing data for a wide range of purposes including forecasting stock and cryptocurrency prices, economy insights, footfall and traffic. A project from


Next Page: 10000

Site Map 2018_01_14
Site Map 2018_01_15
Site Map 2018_01_16
Site Map 2018_01_17
Site Map 2018_01_18
Site Map 2018_01_19
Site Map 2018_01_20
Site Map 2018_01_21
Site Map 2018_01_22
Site Map 2018_01_23
Site Map 2018_01_24
Site Map 2018_01_25
Site Map 2018_01_26
Site Map 2018_01_27
Site Map 2018_01_28
Site Map 2018_01_29
Site Map 2018_01_30
Site Map 2018_01_31
Site Map 2018_02_01
Site Map 2018_02_02
Site Map 2018_02_03
Site Map 2018_02_04
Site Map 2018_02_05
Site Map 2018_02_06
Site Map 2018_02_07
Site Map 2018_02_08
Site Map 2018_02_09
Site Map 2018_02_10
Site Map 2018_02_11
Site Map 2018_02_12
Site Map 2018_02_13
Site Map 2018_02_14
Site Map 2018_02_15
Site Map 2018_02_15
Site Map 2018_02_16
Site Map 2018_02_17
Site Map 2018_02_18
Site Map 2018_02_19
Site Map 2018_02_20
Site Map 2018_02_21
Site Map 2018_02_22
Site Map 2018_02_23
Site Map 2018_02_24
Site Map 2018_02_25
Site Map 2018_02_26
Site Map 2018_02_27
Site Map 2018_02_28
Site Map 2018_03_01
Site Map 2018_03_02
Site Map 2018_03_03
Site Map 2018_03_04
Site Map 2018_03_05
Site Map 2018_03_06
Site Map 2018_03_07
Site Map 2018_03_08
Site Map 2018_03_09
Site Map 2018_03_10
Site Map 2018_03_11
Site Map 2018_03_12
Site Map 2018_03_13
Site Map 2018_03_14
Site Map 2018_03_15
Site Map 2018_03_16
Site Map 2018_03_17
Site Map 2018_03_18
Site Map 2018_03_19
Site Map 2018_03_20
Site Map 2018_03_21
Site Map 2018_03_22
Site Map 2018_03_23
Site Map 2018_03_24
Site Map 2018_03_25
Site Map 2018_03_26
Site Map 2018_03_27
Site Map 2018_03_28
Site Map 2018_03_29
Site Map 2018_03_30
Site Map 2018_03_31
Site Map 2018_04_01
Site Map 2018_04_02
Site Map 2018_04_03
Site Map 2018_04_04
Site Map 2018_04_05
Site Map 2018_04_06
Site Map 2018_04_07
Site Map 2018_04_08
Site Map 2018_04_09
Site Map 2018_04_10
Site Map 2018_04_11
Site Map 2018_04_12
Site Map 2018_04_13
Site Map 2018_04_14
Site Map 2018_04_15
Site Map 2018_04_16
Site Map 2018_04_17
Site Map 2018_04_18
Site Map 2018_04_19
Site Map 2018_04_20
Site Map 2018_04_21
Site Map 2018_04_22
Site Map 2018_04_23
Site Map 2018_04_24
Site Map 2018_04_25
Site Map 2018_04_26
Site Map 2018_04_27
Site Map 2018_04_28
Site Map 2018_04_29
Site Map 2018_04_30
Site Map 2018_05_01
Site Map 2018_05_02
Site Map 2018_05_03
Site Map 2018_05_04
Site Map 2018_05_05
Site Map 2018_05_06
Site Map 2018_05_07
Site Map 2018_05_08
Site Map 2018_05_09
Site Map 2018_05_15
Site Map 2018_05_16
Site Map 2018_05_17
Site Map 2018_05_18
Site Map 2018_05_19
Site Map 2018_05_20
Site Map 2018_05_21
Site Map 2018_05_22
Site Map 2018_05_23
Site Map 2018_05_24
Site Map 2018_05_25
Site Map 2018_05_26
Site Map 2018_05_27
Site Map 2018_05_28
Site Map 2018_05_29
Site Map 2018_05_30
Site Map 2018_05_31
Site Map 2018_06_01
Site Map 2018_06_02
Site Map 2018_06_03
Site Map 2018_06_04
Site Map 2018_06_05
Site Map 2018_06_06
Site Map 2018_06_07
Site Map 2018_06_08
Site Map 2018_06_09
Site Map 2018_06_10
Site Map 2018_06_11
Site Map 2018_06_12
Site Map 2018_06_13
Site Map 2018_06_14
Site Map 2018_06_15
Site Map 2018_06_16
Site Map 2018_06_17
Site Map 2018_06_18
Site Map 2018_06_19
Site Map 2018_06_20
Site Map 2018_06_21
Site Map 2018_06_22
Site Map 2018_06_23
Site Map 2018_06_24
Site Map 2018_06_25
Site Map 2018_06_26
Site Map 2018_06_27
Site Map 2018_06_28
Site Map 2018_06_29
Site Map 2018_06_30
Site Map 2018_07_01
Site Map 2018_07_02
Site Map 2018_07_03
Site Map 2018_07_04
Site Map 2018_07_05
Site Map 2018_07_06
Site Map 2018_07_07
Site Map 2018_07_08
Site Map 2018_07_09
Site Map 2018_07_10
Site Map 2018_07_11
Site Map 2018_07_12
Site Map 2018_07_13
Site Map 2018_07_14
Site Map 2018_07_15
Site Map 2018_07_16
Site Map 2018_07_17
Site Map 2018_07_18
Site Map 2018_07_19
Site Map 2018_07_20
Site Map 2018_07_21
Site Map 2018_07_22
Site Map 2018_07_23
Site Map 2018_07_24
Site Map 2018_07_25
Site Map 2018_07_26
Site Map 2018_07_27
Site Map 2018_07_28
Site Map 2018_07_29
Site Map 2018_07_30
Site Map 2018_07_31
Site Map 2018_08_01
Site Map 2018_08_02
Site Map 2018_08_03
Site Map 2018_08_04
Site Map 2018_08_05
Site Map 2018_08_06
Site Map 2018_08_07
Site Map 2018_08_08
Site Map 2018_08_09
Site Map 2018_08_10
Site Map 2018_08_11
Site Map 2018_08_12
Site Map 2018_08_13
Site Map 2018_08_15
Site Map 2018_08_16
Site Map 2018_08_17
Site Map 2018_08_18
Site Map 2018_08_19
Site Map 2018_08_20
Site Map 2018_08_21
Site Map 2018_08_22
Site Map 2018_08_23
Site Map 2018_08_24
Site Map 2018_08_25
Site Map 2018_08_26
Site Map 2018_08_27
Site Map 2018_08_28
Site Map 2018_08_29
Site Map 2018_08_30
Site Map 2018_08_31
Site Map 2018_09_01
Site Map 2018_09_02
Site Map 2018_09_03
Site Map 2018_09_04
Site Map 2018_09_05
Site Map 2018_09_06
Site Map 2018_09_07
Site Map 2018_09_08
Site Map 2018_09_09
Site Map 2018_09_10
Site Map 2018_09_11
Site Map 2018_09_12
Site Map 2018_09_13
Site Map 2018_09_14
Site Map 2018_09_15
Site Map 2018_09_16
Site Map 2018_09_17
Site Map 2018_09_18
Site Map 2018_09_19
Site Map 2018_09_20
Site Map 2018_09_21
Site Map 2018_09_23
Site Map 2018_09_24
Site Map 2018_09_25
Site Map 2018_09_26
Site Map 2018_09_27
Site Map 2018_09_28
Site Map 2018_09_29
Site Map 2018_09_30
Site Map 2018_10_01
Site Map 2018_10_02
Site Map 2018_10_03
Site Map 2018_10_04
Site Map 2018_10_05
Site Map 2018_10_06
Site Map 2018_10_07
Site Map 2018_10_08
Site Map 2018_10_09
Site Map 2018_10_10