iOS Dev & AI Enthusiast

[tomaszbaranowicz-blog]

AI Weekly 21 April 2018

Hi! New AI Weekly is here! Once again this week was really great for AI developers, huge amount of new code and libraries appeared, Google Developers published couple great articles on Medium like the one about Classifying text with TensorFlow Estimators, also others contributed with couple interesting tutorials, it’s worth to read introduction to Tensorflow.js. There are also worth reading articles in general section, especially the one about current state of AI industry, of course that’s not all… just enjoy your weekend reading other AI news and don’t forget to share it with your friends

GENERAL

A.I. Researchers Are Making More Than $1 Million, Even at a Nonprofit – One of the poorest-kept secrets in Silicon Valley has been the huge salaries and bonuses that experts in artificial intelligence can command. Now, a little-noticed tax filing by a research lab called OpenAI has made some of those eye-popping figures public. OpenAI paid its top researcher, Ilya Sutskever, more than $1.9 million in 2016. It paid another leading researcher, Ian Goodfellow, more than $800,000 — even though he was not hired until March of that year. Both were recruited from Google. https://nyti.ms/2Hi6c4P

Artificial Intelligence — The Revolution Hasn’t Happened Yet – Artificial Intelligence (AI) is the mantra of the current era. The phrase is intoned by technologists, academicians, journalists and venture capitalists alike. As with many phrases that cross over from technical academic fields into general circulation, there is significant misunderstanding accompanying the use of the phrase. But this is not the classical case of the public not understanding the scientists — here the scientists are often as befuddled as the public. The idea that our era is somehow seeing the emergence of an intelligence in silicon that rivals our own entertains all of us — enthralling us and frightening us in equal measure. And, unfortunately, it distracts us. http://bit.ly/2HD1tKB

PROGRAMMING

A Gentle Introduction to TensorFlow.js – Tensorflow.js is a library built on deeplearn.js to create deep learning modules directly on the browser. Using that you can create CNNs, RNNs , etc … on the browser and train these modules using the client’s GPU processing power. Hence, a server GPU is not needed to train the NN. This tutorial starts by explaining the basic building blocks of TensorFlow.js and the operations on them. Then, author describes how to create some complicated models. http://bit.ly/2HOpOuw

Hallucinogenic Deep Reinforcement Learning Using Python and Keras – Teaching a machine to master car racing and fireball avoidance through “World Models” http://bit.ly/2HiDEIG

How to implement a YOLO (v3) object detector from scratch in PyTorch – author uses PyTorch to implement an object detector based on YOLO v3, one of the faster object detection algorithms out there. The code for this tutorial is designed to run on Python 3.5, and PyTorch 0.3. http://bit.ly/2K2jNv7

Building an Iris classifier with eager execution – One of the new additions to TensorFlow in the last months has been the eager execution, an additional low-level interface promising to make development a lot simpler and easier to debug. Whenever eager execution is enabled, operations are executed immediately, instead of having to go through a separate execution step. In a nutshell, this means that writing TF code can be (potentially) as simple as writing pure NumPy code! http://bit.ly/2HI5BZW

Classifying text with TensorFlow Estimators – Throughout this post we will show you how to classify text using Estimators in TensorFlow. Here’s the outline of what it will cover: Loading data using Datasets. Building baselines using pre-canned estimators. Using word embeddings. Building custom estimators with convolution and LSTM layers. Loading pre-trained word vectors. Evaluating and comparing models using TensorBoard. http://bit.ly/2qNYCVo

PAPERS

Evolved Policy Gradients – OpenAI released an experimental metalearning approach called Evolved Policy Gradients, a method that evolves the loss function of learning agents, which can enable fast training on novel tasks. Agents trained with EPG can succeed at basic tasks at test time that were outside their training regime, like learning to navigate to an object on a different side of the room from where it was placed during training. http://bit.ly/2qPKquh

COMPETITION

CVPR 2018 On-Device Visual Intelligence Challenge – public competition for real-time image classification that uses state-of-the-art Google technology to significantly lower the barrier to entry for mobile development. OVIC provides two key features to catalyze innovation: a unified latency metric and an evaluation platform, deadline on June 15th. http://bit.ly/2HH5Owm

Be always on time with weekly: http://theaigeek.com/ai-weekly-21-april-2018/ 

[tomaszbaranowicz-blog]

AI Weekly 13 April 2018

Hi! New AI Weekly is here! This week was really great for AI developers, huge amount of new code and libraries appeared, lot of them published during TF Dev Summit. There are also worth reading articles in learning section, especially the one with notes about what might surprise you when trying to reproduce paper. Talking about papers, just don’t miss the one explaining usage of GAN for image compression, it’s awesome, of course that’s not all... just enjoy your weekend reading other AI news and don’t forget to share it with your friends ;) 

LEARNING 

Differentiable Plasticity: A New Method for Learning to Learn - biological brains exhibit plasticity—that is, the ability for connections between neurons to change continually and automatically throughout life, allowing animals to learn quickly and efficiently from ongoing experience. The levels of plasticity of different areas and connections in the brain are the result of millions of years of fine-tuning by evolution to allow efficient learning during the animal’s lifetime. The resultant ability to learn continually over life lets animals adapt to changing or unpredictable environments with very little additional data. https://ubr.to/2qt9MyB

Lessons Learned Reproducing a Deep Reinforcement Learning Paper - reproducing papers is a good way of levelling up machine learning skills, if you’re thinking about reproducing papers too, here are some notes on what surprised me about working with deep RL. http://bit.ly/2qsZxti

VIDEO 

Heroes of Deep Learning: Andrew Ng interviews Yann LeCun http://bit.ly/2EJTUMu

PROGRAMMING 

Magenta - browser-based applications, many of which are implemented with TensorFlow.js for WebGL-accelerated inference. http://bit.ly/2GYUdFh

Interactive supervision with TensorBoard - IBM Research AI implemented semi-supervision in TensorBoard t-SNE and contributed components required for interactive supervision to demonstrate cognitive-assisted labeling. A metadata editor, distance metric/space selection, neighborhood function selection, and t-SNE perturbation were added to TensorBoard in addition to semi-supervision for t-SNE. These components function in concert to apply a partial labeling that informs semi-supervised t-SNE to clarify the embedding and progressively ease the labeling burden. https://ibm.co/2ITnIsN

Fitting larger networks into memory - the python/Tensorflow package openai/gradient-checkpointing, that lets you fit 10x larger neural nets into memory at the cost of an additional 20% computation time. http://bit.ly/2IOyrVh

TensorFlow Probability - a probabilistic programming toolbox for machine learning researchers and practitioners to quickly and reliably build sophisticated models that leverage state-of-the-art hardware. http://bit.ly/2GYRVpu

Python Script to download hundreds of images from 'Google Images'. It is a ready-to-run code! http://bit.ly/2veI2Tm

PAPERS 

Looking to Listen: Audio-Visual Speech Separation - people are remarkably good at focusing their attention on a particular person in a noisy environment, mentally “muting” all other voices and sounds. Known as the cocktail party effect, this capability comes natural to us humans. However, automatic speech separation — separating an audio signal into its individual speech sources — while a well-studied problem, remains a significant challenge for computers. http://bit.ly/2HwNnYE

Towards a Virtual Stuntman - Motion control problems have become standard benchmarks for reinforcement learning, and deep RL methods have been shown to be effective for a diverse suite of tasks ranging from manipulation to locomotion. However, characters trained with deep RL often exhibit unnatural behaviours, bearing artifacts such as jittering, asymmetric gaits, and excessive movement of limbs. Can we train our characters to produce more natural behaviours? http://bit.ly/2GW1IRp

Generative Adversarial Networks for Extreme Learned Image Compression - a framework for extreme learned image compression based on Generative Adversarial Networks (GANs), obtaining visually pleasing images at significantly lower bitrates than previous methods. This is made possible through our GAN formulation of learned compression combined with a generator/decoder which operates on the full-resolution image and is trained in combination with a multi-scale discriminator. Additionally, this method can fully synthesize unimportant regions in the decoded image such as streets and trees from a semantic label map extracted from the original image, therefore only requiring the storage of the preserved region and the semantic label map. A user study confirms that for low bitrates, this approach significantly outperforms state-of-the-art methods, saving up to 67% compared to the next-best method BPG. http://bit.ly/2qq2nQP

RESOURCES 

EPIC-Kitchens - The largest dataset in first-person (egocentric) vision; multi-faceted non-scripted recordings in native environments - i.e. the wearers' homes, capturing all daily activities in the kitchen over multiple days. Annotations are collected using a novel `live' audio commentary approach. 55 hours of recording - Full HD, 60fps, 11.5M frames http://bit.ly/2GWn55d

[tomaszbaranowicz-blog]

AI Weekly 6 April 2018

Hi! New AI Weekly is here! This week was really great for AI industry, at first it’s definitely worth to watch all videos from TensorFlow Dev Summit 2018. From general news, Apple has hired Google’s chief of search and artificial intelligence, John Giannandrea and Google employees have signed a letter protesting the company’s involvement in a Pentagon program. There are also 2 interesting papers, which were published, especially the one about The Tsetlin Machine, which seems to be interesting alternative for neural networks. If you are developer i strongly recommend taking part in OpenAI Retro Contest, which will be really fun, of course that’s not all... just enjoy your weekend reading other AI news and don’t forget to share it with your friends ;) 

GENERAL Google Workers Urge C.E.O. to Pull Out of Pentagon A.I. Project - Thousands of Google employees, including dozens of senior engineers, have signed a letter protesting the company’s involvement in a Pentagon program that uses artificial intelligence to interpret video imagery and could be used to improve the targeting of drone strikes. The letter, which is circulating inside Google and has garnered more than 3,100 signatures, reflects a culture clash between Silicon Valley and the federal government that is likely to intensify as cutting-edge artificial intelligence is increasingly employed for military purposes. https://nyti.ms/2Ivl9gk

Apple Hires Google’s A.I. Chief - Apple has hired Google’s chief of search and artificial intelligence, John Giannandrea, a major coup in its bid to catch up to the artificial intelligence technology of its rivals. Apple said on Tuesday that Mr. Giannandrea will run Apple’s “machine learning and A.I. strategy,” and become one of 16 executives who report directly to Apple’s chief executive, Timothy D. Cook. The hire is a victory for Apple, which many Silicon Valley executives and analysts view as lagging its peers in artificial intelligence, an increasingly crucial technology for companies that enable computers to handle more complex tasks, like understanding voice commands or identifying people in images. https://nyti.ms/2qaCRio

LEARNING Course Project Reports for 2017 CS224n: Natural Language Processing with Deep Learning https://stanford.io/2q73GEc

VIDEO TensorFlow Dev Summit 2018 videos http://bit.ly/2GG4GcM

An introduction to Reinforcement Learning http://bit.ly/2HdBwyq

MIT AGI: Boston Dynamics (Marc Raibert, CEO) http://bit.ly/2qfqHEe

PROGRAMMING MobileNetV2 - the next generation of on-device models that push the state of the art for mobile visual recognition including classification, object detection and semantic segmentation. http://bit.ly/2Eq0iZa

CometML wants to do for machine learning what GitHub did for code - Comet.ml allows data scientists and developers to easily monitor, compare and optimize their machine learning models. The service provides you with a dashboard that brings together the code of your machine learning (ML) experiments and their results. In addition, the service also allows you to optimize your models by tweaking the hyperparameters of your experiments. As you train your model, Comet tracks the results and provides you with a graph of your results, but it also tracks your code changes and imports them so that you can later compare all the different aspects of the various versions of your experiments. https://tcrn.ch/2Hdj9tq

PAPERS The Tsetlin Machine - Although simple individually, artificial neurons provide state-of-the-art performance when interconnected in deep networks. Unknown to many, there exists an arguably even simpler and more versatile learning mechanism, namely, the Tsetlin Automaton. Merely by means of a single integer as memory, it learns the optimal action in stochastic environments. In this paper, author introduces the Tsetlin Machine, which solves complex pattern recognition problems with easy-to-interpret propositional formulas, composed by a collective of Tsetlin Automata. http://bit.ly/2JqQGB1

Learning to navigate in cities without a map - in this paper DeepMind present an interactive navigation environment that uses first-person perspective photographs from Google Street View and gamify that environment to train an AI. As standard with Street View images, faces and license plates have been blurred and are unrecognisable. They build a neural network-based artificial agent that learns to navigate multiple cities using visual information (pixels from a Street View image). http://bit.ly/2H4EEiz

CONTESTS OpenAI Retro Contest - April 5 to June 5, 2018 - transfer-learning contest using the Sonic The Hedgehog™ series of games for SEGA Genesis. In this contest, participants try to create the best agent for playing custom levels of the Sonic games — without have access to those levels during development. http://bit.ly/2qcCBi6 If you want to be always on time with AI weekly, feel free to subscribe to newsletter on http://theaigeek.com/

[tomaszbaranowicz-blog]

AI Weekly 30 Mar 2018

Hi! New AI Weekly is here! This week brought us only a couple significant news, but hey TensorFlow Dev Summit 2018 just started, so next weekly will be full of awesome news for developers, i’m sure 😉 Going back to past week, Google released Cloud Text-to-Speech, which now allows everyone to use text-to-speech synthesis with DeepMind WaveNet technology, check it out as the quality is impressive. DeepMind published paper explaining how they tried to learn artificial agents to generate images without any labeled data. In this weekly you will also find a great article about Monte Carlo Tree Search and a lot more... just enjoy your weekend reading other AI news and don’t forget to share it with your friends ;) 

GENERAL AI can spot signs of Alzheimer’s before your family does - Earlier diagnosis could help researchers develop drugs to slow the progress of the disease. http://bit.ly/2pTmiqQ

What worries me about AI by François Chollet - author of Keras http://bit.ly/2GoxA0w

LEARNING An overview of image classification networks - Learning about the different network architectures for image classification is a daunting task. In this blog post author discusses the main architectures that are currently available in the keras package. Hel goes through these architectures in a chronological order and attempt to discuss their advantages and disadvantages from the perspective of a practitioner. http://bit.ly/2uxX4mG

Monte Carlo Tree Search – beginners guide http://bit.ly/2H2BNEB

PROGRAMMING TensorFlow Dev Summit 2018 - this year event starts today (30th of March) and all videos will be available on new dedicated youtube channel, must see for all AI/ML developers: http://bit.ly/2GnzVsU

Google Introducing Cloud Text-to-Speech - Many Google products come with built-in high-quality text-to-speech synthesis that produces natural sounding speech. Now they're bringing this technology to Google Cloud Platform with Cloud Text-to-Speech. Cloud Text-to-Speech lets you choose from 32 different voices from 12 languages and variants. Cloud Text-to-Speech correctly pronounces complex text such as names, dates, times and addresses for authentic sounding speech right out of the gate. Cloud Text-to-Speech also allows you to customize pitch, speaking rate, and volume gain, and supports a variety of audio formats, including MP3 and WAV. http://bit.ly/2GlO4qu

TensorFlow 1.7.0 released http://bit.ly/2Ik2CUa

PAPERS Learning to write programs that generate images - DeepMind equipped artificial agents with the same tools that we use to generate images and demonstrate that they can reason about how digits, characters and portraits are constructed. Crucially, they learn to do this by themselves and without the need for human-labelled datasets. This contrasts with recent research which has so far relied on learning from human demonstrations, which can be a time-intensive process. http://bit.ly/2E74hKc

Using Machine Learning to Discover Neural Network Optimizers - Deep learning models have been deployed in numerous Google products, such as Search, Translateand Photos. The choice of optimization method plays a major role when training deep learning models. For example, stochastic gradient descent works well in many situations, but more advanced optimizers can be faster, especially for training very deep networks. Coming up with new optimizers for neural networks, however, is challenging due to to the non-convex nature of the optimization problem. On the Google Brain team, they wanted to see if it could be possible to automate the discovery of new optimizers, in a way that is similar to how AutoML has been used to discover new competitive neural network architectures. http://bit.ly/2GGx4uh

[tomaszbaranowicz-blog]

AI Weekly 23 Mar 2018

Hi! New AI Weekly is here! This week brought us sad news about first victim of autonomous car. On the other hand Baidu released huge self-driving dataset to let others work on autonomous driving technology and improve it. DARPA announced that they want to use AI to design new parts for military. This week was also great because of 2 significant papers which were published, one comparing random search with reinforcement learning and second explaining how to understand neural network by neuron deletion. Of course that’s not all… just enjoy your weekend reading other AI news and don’t forget to share it with your friends ;) 

GENERAL How the AI cloud could produce the richest companies ever - Amazon, Google, and Microsoft all want to dominate the business of providing artificial-intelligence services through cloud computing. The winner may have the OS of the future. http://bit.ly/2HZzNMk

The US military wants AI to dream up weird new helicopters - That’s the goal of a new project from the Defense Advanced Research Projects Agency (DARPA), the research wing of the US Defense Department. DARPA wants entrants to rethink the way complex components are designed by combining recent advances in machine learning with fundamental tenets of math and engineering. http://bit.ly/2Giygnr

Self-driving Uber kills Arizona woman in first fatal crash involving pedestrian - An autonomous Uber car killed a woman in the street in Arizona, police said, in what appears to be the first reported fatal crash involving a self-driving vehicle and a pedestrian in the US. Tempe police said the self-driving car was in autonomous mode at the time of the crash and that the vehicle hit a woman, who was walking outside of the crosswalk and later died at a hospital. There was a vehicle operator inside the car at the time of the crash. http://bit.ly/2pwuvkP

Scientists taught an AI system to diagnose brain cancer - A new research effort by an international team of scientists reveals that machine-learning algorithms can be a powerful tool for medicine. The group, which published its work in the journal Nature, managed to create and train an AI to successfully identify different types of brain tumors with impressive accuracy. https://nyp.st/2IKjxjs

VIDEOS MIT AGI: Cognitive Architecture (Nate Derbinsky) http://bit.ly/2GfFlVZ

RESOURCES Baidu Apollo Releases Massive Self-driving Dataset - Baidu this Thursday announced the release of ApolloScape, billed as the world’s largest open-source dataset for autonomous driving technology. ApolloScape was released under Baidu’s autonomous driving platform Apollo, which Baidu hopes will become “the Android of the auto industry.” Apollo gives developers access to a complete set of service solutions and open-source codes and can enable for example a software engineer to convert a Lincoln MKZ into a self-driving vehicle in about 48 hours. ApolloScape’s open sourced data now provides developers a base for building self-driving vehicles. http://bit.ly/2G8lCnB

PROGRAMMING Live training loss plot in Jupyter Notebook for Keras, PyTorch and others http://bit.ly/2pAbJJ8

GAN with Keras: Application to Image Deblurring - In 2014, Ian Goodfellow introduced the Generative Adversarial Networks(GAN). This article focuses on applying GAN to Image Deblurring withKeras. http://bit.ly/2G28dlm

PAPERS Simple random search provides a competitive approach to reinforcement learning - A common belief in model-free reinforcement learning is that methods based on random search in the parameter space of policies exhibit significantly worse sample complexity than those that explore the space of actions. Authors dispel such beliefs by introducing a random search method for training static, linear policies for continuous control problems, matching state-of-the-art sample efficiency on the benchmark MuJoCo locomotion tasks. http://bit.ly/2G0MASt

Understanding deep learning through neuron deletion by DeepMind - deep neural networks are composed of many individual neurons, which combine in complex and counterintuitive ways to solve a wide range of challenging tasks. This complexity grants neural networks their power but also earns them their reputation as confusing and opaque black boxes. Understanding how deep neural networks function is critical for explaining their decisions and enabling us to build more powerful systems. For instance, imagine the difficulty of trying to build a clock without understanding how individual gears fit together. One approach to understanding neural networks, both in neuroscience and deep learning, is to investigate the role of individual neurons, especially those which are easily interpretable. http://bit.ly/2INEGJB

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 16 Mar 2018

Hi! New AI Weekly is here! This week was really good and brought us new Developer Survey Results, with chapter devoted to AI, it’s definitely worth reading, also Microsoft claimed that they have created Chinese to English translator with the same quality and accuracy as a person, which would be a breakthrough achievement. Of course there are great articles from Google Research, one about Evolutionary AutoML to Discover Neural Network Architectures and second about technology behind the Motion Photos in Pixel 2. That was also great week for programmers, but to not make this introduction longer i will only suggest to check this section if you want to find out how faceID in iPhone X works… ok enough of intro, just enjoy your weekend reading other AI news and don’t forget to share it with your friends ;) 

GENERAL Developer Survey Results 2018 - over 100,000 developers took the 30-minute survey this past January, there is a great part called "Technology and Society” devoted mostly to AI, really worth reading. http://bit.ly/2FF9rCI

Microsoft reaches a historic milestone, using AI to match human performance in translating news from Chinese to English - A team of Microsoft researchers said Wednesday that they believe they have created the first machine translation system that can translate sentences of news articles from Chinese to English with the same quality and accuracy as a person. http://bit.ly/2Gym3sx

VIDEOS MIT Self-Driving Cars: Sterling Anderson, Co-Founder, Aurora http://bit.ly/2tPRrQu 

LEARNING Behind the Motion Photos Technology in Pixel 2 - One of the most compelling things about smartphones today is the ability to capture a moment on the fly. With motion photos, a new camera feature available on the Pixel 2 and Pixel 2 XL phones, you no longer have to choose between a photo and a video so every photo you take captures more of the moment. When you take a photo with motion enabled, your phone also records and trims up to 3 seconds of video. Let’s take a look behind the technology that makes this possible! http://bit.ly/2IxSY16

Using Evolutionary AutoML to Discover Neural Network Architectures - The brain has evolved over a long time, from very simple worm brains 500 million years ago to a diversity of modern structures today. The human brain, for example, can accomplish a wide variety of activities, many of them effortlessly — telling whether a visual scene contains animals or buildings feels trivial to us, for example. To perform activities like these, artificial neural networks require careful design by experts over years of difficult research, and typically address one specific task, such as to find what's in a photograph, to call a genetic variant, or to help diagnose a disease. Ideally, one would want to have an automated method to generate the right architecture for any given task. One approach to generate these architectures is through the use of evolutionary algorithms. http://bit.ly/2FHPStq

PROGRAMMING MusicVAE: Creating a palette for musical scores with machine learning. - When a painter creates a work of art, she first blends and explores color options on an artist’s palette before applying them to the canvas. This process is a creative act in its own right and has a profound effect on the final work. Musicians and composers have mostly lacked a similar device for exploring and mixing musical ideas, but Google is hoping to change that. They have introduced MusicVAE, a machine learning model that lets you create palettes for blending and exploring musical scores. http://bit.ly/2IvBgeD

How to implement iPhone X’s FaceID using Deep Learning in Python - Reverse engineering iPhone X’s new unlocking mechanism http://bit.ly/2pekTuT

Comparing Deep Learning Frameworks: A Rosetta Stone Approach - Microsoft developers believe deep-learning frameworks are like languages: Sure, many people speak English, but each language serves its own purpose. They have created common code for several different network structures and executed it across many different frameworks. Idea was to a create a Rosetta Stone of deep-learning frameworks – assuming you know one well, to help anyone leverage any framework. Situations may arise where a paper publishes code in another framework or the whole pipeline is in another language. Instead of writing a model from scratch in your favourite framework it may be easier to just use the “foreign” language. http://bit.ly/2GzPXfV

Plain python implementations of basic machine learning algorithms - This repository contains implementations of basic machine learning algorithms in plain Python (Python Version 3.6+). All algorithms are implemented from scratch without using additional machine learning libraries. http://bit.ly/2Ix4E44

Google Colab - Train Your Machine Learning Models on Google’s GPUs for Free  http://bit.ly/2DyFVse

PAPERS Independently Recurrent Neural Network (IndRNN): Building A Longer and Deeper RNN - Recurrent neural networks (RNNs) have been widely used for processing sequential data. However, RNNs are commonly difficult to train due to the well-known gradient vanishing and exploding problems and hard to learn long-term patterns. Long short-term memory (LSTM) and gated recurrent unit (GRU) were developed to address these problems, but the use of hyperbolic tangent and the sigmoid action functions results in gradient decay over layers. Consequently, construction of an efficiently trainable deep network is challenging. In addition, all the neurons in an RNN layer are entangled together and their behaviour is hard to interpret. To address these problems, a new type of RNN, referred to as independently recurrent neural network (IndRNN), is proposed in this paper, where neurons in the same layer are independent of each other and they are connected across layers. Researchers have shown that an IndRNN can be easily regulated to prevent the gradient exploding and vanishing problems while allowing the network to learn long-term dependencies. http://bit.ly/2FO3fV8

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 9 Mar 2018

Hi! New AI Weekly is here! This week brought us great articles about use of AI in Healthcare and in programming, there is also awesome explanation about what and how NN understands it’s input data. For prof. Andrew Ng fans there are also great notes from Deep Learning courses… wow that was really cool week, but that’s not all of course… just enjoy your weekend reading other AI news and don’t forget to share it with your friends ;)

GENERAL

Artificial Intelligence and Machine Learning for Healthcare - Artificial Intelligence (AI), machine learning, and deep learning are taking the healthcare industry by storm. They are not pie in the sky technologies any longer; they are practical tools that can help companies optimize their service provision, improve the standard of care, generate more revenue, and decrease risk. Nearly all major companies in the healthcare space have already begun to use the technology in practice; in this post author presents some of the important highlights of the implementation, and what they mean for other companies in healthcare. http://bit.ly/2FoCNVH

Inside the Chinese lab that plans to rewire the world with AI - Alibaba is investing huge sums in AI research and resources—and it is building tools to challenge Google and Amazon. http://bit.ly/2p2mumF

Ubisoft is using AI to catch bugs in games before devs make them - The gaming company's Commit Assistant AI tool has been trained to spot when programmers are about to make a mistake http://bit.ly/2p0A8GE

VIDEOS

Reinforcement Learning: Hidden Theory and New Super-Fast Algorithms - by Sean Meyn, University of Florida http://bit.ly/2FwhGNl

LEARNING

The Building Blocks of Interpretability - With the growing success of neural networks, there is a corresponding need to be able to explain their decisions — including building confidence about how they will behave in the real-world, detecting model bias, and for scientific curiosity. In order to do so, we need to both construct deep abstractions and reify (or instantiate) them in rich interfaces . With a few exceptions , existing work on interpretability fails to do these in concert. http://bit.ly/2Ifnv3i

RESOURCES

Notes from Coursera Deep Learning courses by Andrew Ng - http://bit.ly/2oXWJop

PROGRAMMING

Open Sourcing the Hunt for Exoplanets - Google released their code for processing the Kepler data, training neural network model, and making predictions about new candidate signals. http://bit.ly/2Icb0pk

Tensor Comprehensions - Facebook AI Research (FAIR) announced the release of Tensor Comprehensions, a C++ library and mathematical language that helps bridge the gap between researchers, who communicate in terms of mathematical operations, and engineers focusing on the practical needs of running large-scale models on various hardware backends. The main differentiating feature of Tensor Comprehensions is that it represents a unique take on Just-In-Time compilation to produce the high-performance codes that the machine learning community needs, automatically and on-demand. http://bit.ly/2DdVCVN

PAPERS

Reptile: A Scalable Meta-Learning Algorithm - OpenAI developed a simple meta-learning algorithm called Reptile which works by repeatedly sampling a task, performing stochastic gradient descent on it, and updating the initial parameters towards the final parameters learned on that task. This method performs as well as MAML, a broadly applicable meta-learning algorithm, while being simpler to implement and more computationally efficient. http://bit.ly/2txfkfy

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 2 Mar 2018

Hi! New AI Weekly is here! This week brought us couple great articles, definitely worth reading is the first from series about Harvard progress in AI, their robots are really amazing. Also worth to mention that Google announced online Machine Learning Course, it looks really promising and is must check this week. That’s not all of course… just enjoy your weekend reading other AI news and don’t forget to share it with your friends ;) 

GENERAL An AI just beat top lawyers at their own game - The nation's top lawyers recently battled artificial intelligence in a competition to interpret contracts — and they lost. A new study, conducted by legal AI platform LawGeex in consultation with law professors from Stanford University, Duke University School of Law, and University of Southern California, pitted twenty experienced lawyers against an AI trained to evaluate legal contracts. Competitors were given four hours to review five non-disclosure agreements (NDAs) and identify 30 legal issues, including arbitration, confidentiality of relationship, and indemnification. They were scored by how accurately they identified each issue. http://on.mash.to/2oCX3Zw

Behind the Chat: How E-commerce Robot Assistant AliMe Works http://bit.ly/2F6t2Y9

Onward and upward, robots - Harvard scientists help drive new age of machines, aiming for transformative impact in medicine, on Main Street, and beyond. This is first in a series of articles on cutting-edge research at Harvard http://bit.ly/2FhJGY9

Artificial Intelligence - What it is and why it matters by SAS http://bit.ly/2FM8K7r

VIDEOS Variational Autoencoders - In this episode, Arxiv Insights dives into Variational Autoencoders, a class of neural networks that can learn to compress data completely unsupervised! http://bit.ly/2FLGXnJ

LEARNING Machine Learning Crash Course by Google - A self-study guide for aspiring machine learning practitioners. Learn Machine Learning concepts by taking the same course that over 10,000 Google engineers have completed (available in English, Spanish, French, Korean, and Mandarin). http://bit.ly/2FLldbu

RESOURCES Google-Landmarks: A New Dataset and Challenge for Landmark Recognition - a new dataset for landmark recognition containing 2M+ images depicting 30K unique landmarks from across the world. http://bit.ly/2tcVwxW

PROGRAMMING Multi-Agent Deep Deterministic Policy Gradient (MADDPG) - This is the code for implementing the MADDPG algorithm presented in the paper: Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments. It is configured to be run in conjunction with environments from the Multi-Agent Particle Environments (MPE). http://bit.ly/2H194yH

Ingredients for Robotics Research - OpenAI released eight simulated robotics environments and a Baselines implementation of Hindsight Experience Replay, all developed for their research over the past year. They’ve used these environments to train models which work on physical robots. They’ve also released a set of requests for robotics research. http://bit.ly/2CTkJgo

PAPERS Learning by playing - DeepMind proposes a new learning paradigm called ‘Scheduled Auxiliary Control (SAC-X)’ which seeks to overcome this exploration issue. SAC-X is based on the idea that to learn complex tasks from scratch, an agent has to learn to explore and master a set of basic skills first. Just as a baby must develop coordination and balance before she crawls or walks—providing an agent with internal (auxiliary) goals corresponding to simple skills increases the chance it can understand and perform more complicated tasks. http://bit.ly/2oCnMW2

Machine Theory of Mind - Theory of mind (ToM; Premack & Woodruff, 1978) broadly refers to humans' ability to represent the mental states of others, including their desires, beliefs, and intentions. DeepMind team proposes to train a machine to build such models too. They design a Theory of Mind neural network -- a ToMnet -- which uses meta-learning to build models of the agents it encounters, from observations of their behaviour alone. Through this process, it acquires a strong prior model for agents' behaviour, as well as the ability to bootstrap to richer predictions about agents' characteristics and mental states using only a small number of behavioural observations. http://bit.ly/2te4paB If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 23 Feb 2018

Hi! New AI Weekly is here! This week brought us couple very interesting papers from Baidu, Google and OpenAI, all of them are really worth reading! There are also cool videos to watch, especially the series published by Institute for Pure & Applied Mathematics provides lot of useful knowledge. That’s not all of course… just enjoy your weekend reading other AI news and don’t forget to share it with your friends ;)

GENERAL 10 BREAKTHROUGH TECHNOLOGIES 2018 - Dueling neural networks. Artificial embryos. AI in the cloud. MIT Technology Review prepared annual list of the 10 technology advances they think will shape the way we work and live now and for years to come. http://bit.ly/2CFU9r4

A Dozen Times Artificial Intelligence Startled The World - The best uses of Generative Models and how they work. http://bit.ly/2FpYO39

Google’s new AI algorithm predicts heart disease by looking at your eyes - Scientists from Google and its health-tech subsidiary Verily have discovered a new way to assess a person’s risk of heart disease using machine learning. By analyzing scans of the back of a patient’s eye, the company’s software is able to accurately deduce data, including an individual’s age, blood pressure, and whether or not they smoke. This can then be used to predict their risk of suffering a major cardiac event — such as a heart attack — with roughly the same accuracy as current leading methods. http://bit.ly/2HGar6X

VIDEOS New Deep Learning Techniques (Lectures) - Institute for Pure & Applied Mathematics http://bit.ly/2sQo2Fm

2018 Isaac Asimov Memorial Debate: Artificial Intelligence - Isaac Asimov’s famous Three Laws of Robotics might be seen as early safeguards for our reliance on artificial intelligence, but as Alexa guides our homes and automated cars replace human drivers, are those Three Laws enough? Neil deGrasse Tyson, Frederick P. Rose Director of the Hayden Planetarium, hosts and moderates a lively discussion about how A.I. is opening doors to limitless possibilities, and if we’re ready for them. The 2018 Isaac Asimov Memorial Debate took place at the Museum on February, 13, 2018. http://bit.ly/2EPMjwI

Visibility and Monitoring for Machine Learning Models - Josh Willis, an engineer at Slack, spoke at our January MeetUp about testing machine learning models in production. http://bit.ly/2ChlrrZ

PROGRAMMING Easily Build a Neural Net for Breast Cancer detection - a simple tutorial to train a neural network by reading data from a CSV http://bit.ly/2sOhuHc

PAPERS The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation - This paper forecasts how malicious actors could misuse AI technology, and potential ways we can prevent and mitigate these threats. http://bit.ly/2CfFLKu

Assessing Cardiovascular Risk Factors with Computer Vision - Using deep learning algorithms trained on data from 284,335 patients, Google Researchers were able to predict CV risk factors from retinal images with surprisingly high accuracy for patients from two independent datasets of 12,026 and 999 patients. http://bit.ly/2CDPiGT

A Closed-form Solution to Photorealistic Image Stylization - Photorealistic image style transfer algorithms aim at stylizing a content photo using the style of a reference photo with the constraint that the stylized photo should remains photorealistic. While several methods exist for this task, they tend to generate spatially inconsistent stylizations with noticeable artifacts. In addition, these methods are computationally expensive, requiring several minutes to stylize a VGA photo. In this paper, authors present a novel algorithm to address the limitations. http://bit.ly/2CeoLnB

Neural Voice Cloning with a Few Samples - In this study, Baidu Research focus on two fundamental approaches for solving the problems with voice cloning: speaker adaptation and speaker encoding. Both techniques can be adapted to a multi-speaker generative speech model with speaker embeddings, without degrading its quality. In terms of naturalness of the speech and similarity to the original speaker, both demonstrate good performance, even with very few cloning audios. http://bit.ly/2GEWBk7

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 16 Feb 2018

Hi! New AI Weekly is here! This week brought us a ton of great learning resources and papers, especially interesting is publication from OpenAI. There is also interesting article about salaries in AI industry. If you are mobile developer, then definitely watch the On-device Learning with Tensorflow and read awesome tutorial about implementing Machine Learning using React Native. That’s not all of course… just enjoy your weekend reading other AI news and don’t forget to share it with your friends ;)

GENERAL Sky-High Salaries Are the Weapons in the AI Talent War - If you want to command a multiyear, seven-figure salary, you used to have only four career options: chief executive officer, banker, celebrity entertainer, or pro athlete. Now there’s a fifth—artificial intelligence expert. One reason: No one can quite agree on how many there are. https://bloom.bg/2o2QQWb

Neural networks everywhere - New chip reduces neural networks’ power consumption by up to 95 percent, making them practical for battery-powered devices. http://bit.ly/2o3yKDw

Stanford AI for Healthcare - great articles about usage of ML in Healthcare http://bit.ly/2oc7KB1

GREEDY, BRITTLE, OPAQUE, AND SHALLOW: THE DOWNSIDES TO DEEP LEARNING - We've been promised a revolution in how and why nearly everything happens. But the limits of modern artificial intelligence are closer than we think. http://bit.ly/2syn6W3

China Will Overtake the U.S. in AI: Alphabet Chairman http://bit.ly/2Bx86uk

VIDEOS On-device Machine Learning With TensorFlow - Machine learning has traditionally been the solely performed on servers and high performance machines. But there is great value is having on-device machine learning for mobile devices. Doing ML inference on mobile devices has huge potential and is still in its early stages. However, it's already more powerful than most realize. In this demo-oriented talk, Yufeng shows some examples of deep learning models used for local prediction on mobile devices. Learn how to use TensorFlow to implement a machine learning model that is tailored to a custom dataset, and start making delightful experiences today! http://bit.ly/2F5QZzA

LEARNING Making Sense of the Bias / Variance Trade-off in (Deep) Reinforcement Learning - What goes into a stable, accurate reinforcement signal? http://bit.ly/2BwV6oz

Background removal with deep learning - Background removal is a task that is quite easy to do manually, or semi manually (Photoshop). However, fully automated background removal is quite a challenging task, this post will let you know more about how to do that using Machine Learning. http://bit.ly/2EvIm4d

How I Shipped a Neural Network on iOS with CoreML, PyTorch, and React Native - blog post http://bit.ly/2Bxef9P

RESOURCES HDR+ burst photography dataset, containing 3500+ "bursts" made up of 28000+ images. http://bit.ly/2EwiENh

Easy Machine Learning - A curated list of free pretrained models and tutorials. Plug state of the art sentiment analysis, object recognition, and more into your project in just a few minutes. http://bit.ly/2CoxMX2

PROGRAMMING Tensorflow Segmentation - For some applications it isn't adequate enough to localize an object with a simple bounding box. For instance, you might want to segment an object region once it is detected. This class of problems is called instance segmentation. Now TensorFlow Object Detection API has been updated with image segmentation! http://bit.ly/2EzjWHe

PyTorch implementation of "Efficient Neural Architecture Search via Parameters Sharing" http://bit.ly/2o3oDPl

PAPERS Deep Reinforcement Learning Doesn't Work Yet - Deep reinforcement learning is surrounded by mountains and mountains of hype. And for good reasons! Reinforcement learning is an incredibly general paradigm, and in principle, a robust and performant RL system should be great at everything. Merging this paradigm with the empirical power of deep learning is an obvious fit. Deep RL is one of the closest things that looks anything like AGI, and that’s the kind of dream that fuels billions of dollars of funding. Unfortunately, it doesn’t really work yet. http://bit.ly/2o88Iy8

ICLR2018 papers sorted by their score http://bit.ly/2EwcV5M

Interpretable Machine Learning through Teaching - OpenAI designed a method that encourages AIs to teach each other with examples that also make sense to humans. This approach automatically selects the most informative examples to teach a concept — for instance, the best images to describe the concept of dogs — and experimentally they found our approach to be effective at teaching both AIs and humans. http://bit.ly/2C3azhm If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 9 Feb 2018

Hi! New AI Weekly is here! This week biggest news was definitely a mind-reading AI, which can see what you're thinking and draw a picture of it. Also lot of awesome papers from OpenAI, DeepMind and others. That’s not all of course, there is also very interesting article about Machine Learning used to engage mobile app users and more... Enjoy your weekend reading other AI news and don’t forget to share it with your friends ;) 

  GENERAL 

Shining Up a Rusty Industry with Artificial Intelligence - One of the primary activities that companies pursue with analytics and data is to plan and optimize operations; this has been a long-term focus of the “operations research” approach to analytics. It has always been done on a relatively small scale, however, using individual models with only a few variables. Cognitive tools—and machine learning in particular—can take this activity to the next level in breadth and depth. http://bit.ly/2H1tacP

This mind-reading AI can see what you're thinking and draw a picture of it - Scientists around the world are racing to be the first to develop artificially intelligent algorithms that can see inside our minds. The idea is not new: in the science fiction of the 1950s and 60s, crazed doctors were frequently seen putting weird contraptions on people’s heads to decipher their thoughts. British TV serial Quatermass and the Pit – in which such a machine is used to translate the thoughts of alien invaders – is a prime example. Now reality is catching up with fantasy. In the past year, AI experts in China, the US and Japan have published research showing that computers can replicate what people are thinking about by using functional magnetic resonance imaging (or fMRI) machines – which measure brain activity – linked to deep neural networks, which replicate human brain functions. http://bit.ly/2nOM1zH

How we grew from 0 to 4 million women on our fashion app, with a vertical machine learning approach - Three years ago launched Chicisimo, it's goal was to offer automated outfit advice. Today, with over 4 million women on the app, developers share how data and machine learning approach helped them grow. http://bit.ly/2ExHX0k

VIDEOS 

MIT AGI: Building machines that see, learn, and think like people (Josh Tenenbaum) - This is a talk by Josh Tenenbaum for course 6.S099: Artificial General Intelligence. This class is free and open to everyone. Our goal is to take an engineering approach to exploring possible paths toward building human-level intelligence for a better world. http://bit.ly/2nO2jc3

LEARNING

Introducing capsule networks - How CapsNets can overcome some shortcomings of CNNs, including requiring less training data, preserving image details, and handling ambiguity. http://oreil.ly/2sonuGE

PAPERS 

Generating Wikipedia by Summarizing Long Sequences - Authors show that generating English Wikipedia articles can be approached as a multi- document summarization of source documents. They use extractive summarization to coarsely identify salient information and a neural abstractive model to generate the article. http://bit.ly/2G0OlKH

DensePose: Dense Human Pose Estimation In The Wild - Dense human pose estimation aims at mapping all human pixels of an RGB image to the 3D surface of the human body. http://bit.ly/2G2r773

IMPALA: Scalable Distributed DeepRL in DMLab-30 - Deep Reinforcement Learning (DeepRL) has achieved remarkable success in a range of tasks, from continuous control problems in robotics to playing games like Go and Atari. The improvements seen in these domains have so far been limited to individual tasks where a separate agent has been tuned and trained for each task. In most recent work, DeepMind explores the challenge of training a single agent on many tasks. http://bit.ly/2sljMgU

Discovering Types for Entity Disambiguation - OpenAI built a system for automatically figuring out which object is meant by a word by having a neural network decide if the word belongs to each of about 100 automatically-discovered “types” (non-exclusive categories). For example, given a sentence like “the prey saw the jaguarcross the jungle”, rather than trying to reason directly whether jaguarmeans the car, the animal, or something else, the system plays “20 questions” with a pre-chosen set of categories. This approach gives a big boost in state-of-the-art on several entity disambiguation datasets. http://bit.ly/2Ef2Uhm 

  If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 2 Feb 2018

Hi! New AI Weekly is here! This week biggest news was definitely official launch of Andrew Ng AI Fund, but that’s not all of course, there is also very interesting paper from DeepMind and more... Enjoy your weekend reading other AI news and don’t forget to share it with your friends ;) 

GENERAL Andrew Ng officially launches his $175M AI Fund - As the founder of the Google Brain deep learning project and co-founder of Coursera, Andrew Ng was one of the most recognizable names in the machine learning community when he became Baidu’s chief scientist in 2014. He left there in early 2017 and quickly launched a number of new AI projects, including the Deeplearning.ai course and Landing.ai, a project that aims to bring AI to manufacturing companies. It turns out that what he was really working on, though, was his AI Fund http://tcrn.ch/2DX9TLJ

VIDEOS Artificial Intelligence vs Machine Learning - Gary explains http://bit.ly/2nFgPlw Neural Networks For Beginners: Create A Neural Network For Wine Classification - Learn how to create a neural network to classify wine in about 15-20 lines of Python, using the Keras library. http://bit.ly/2s5dL7I

LEARNING The Matrix Calculus You Need For Deep Learning - This paper is an attempt to explain all the matrix calculus you need in order to understand the training of deep neural networks. http://bit.ly/2nxWfEN

PAPERS Learning explanatory rules from noisy data - this paper demonstrates it is possible for systems to combine intuitive perceptual with conceptual interpretable reasoning. The system authors describe, ∂ILP, is robust to noise, data-efficient, and produces interpretable rules. http://bit.ly/2s1C6vh

Scalable and accurate deep learning for electronic health records - Predictive modeling with electronic health record (EHR) data is anticipated to drive personalized medicine and improve healthcare quality. Constructing predictive statistical models typically requires extraction of curated predictor variables from normalized EHR data, a labor-intensive process that discards the vast majority of information in each patient's record. Authors propose a representation of patients' entire, raw EHR records based on the Fast Healthcare Interoperability Resources (FHIR) format. They demonstrate that deep learning methods using this representation are capable of accurately predicting multiple medical events from multiple centers without site-specific data harmonization. http://bit.ly/2nDklNt

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 26 Jan 2018

Hi! New AI Weekly is here! This week once again we focused on learning resources. You can read about how twitter is cropping photos using ML, how to solve TSP problem using self-organizing maps, how to build analytics tool for League of Legends… oh, and a lot more. Enjoy your weekend reading other AI news and don’t forget to share it with your friends ;)

GENERAL The head of Facebook’s AI research is stepping into a new role as it shakes up management - Facebook has replaced the man Mark Zuckerberg recruited to run its artificial intelligence research, filling the role with an outsider who will take a bigger role at the company as it puts more AI into its News Feed and other products. http://bit.ly/2GgJMgp

PyTorch, a year in…. - PyTorch was released publicly 1 year ago. Over the last year, we’ve seen an amazing community of people using, contributing to and evangelizing PyTorch. http://bit.ly/2DG1Wup

PROGRAMMING Facebook open sources Detectron - Facebook AI Research (FAIR) open sourced Detectron — our state-of-the-art platform for object detection research. The Detectron project was started in July 2016 with the goal of creating a fast and flexible object detection system built on Caffe2, which was then in early alpha development. Over the last year and a half, the codebase has matured and supported a large number of our projects, including Mask R-CNN and Focal Loss for Dense Object Detection, which won the Marr Prize and Best Student Paper awards, respectively, at ICCV 2017. http://bit.ly/2DFMQoB

Recurrent Neural Networks for Drawing Classification - Train your own Quick, Draw! (https://quickdraw.withgoogle.com/ ) classifier with this new RNN tutorial and public dataset (50M drawings in 345 categories), on the master branch! http://bit.ly/2EbOH1z

DeepLeague: leveraging computer vision and deep learning on the League of Legends mini map - DeepLeague is the first algorithm and dataset (over 100,000 images) that combines computer vision, deep learning, and League of Legends to move LoL analytics to the next level by giving developers easy access to the data encoded in the pixels of the game. http://bit.ly/2BxpOKQ

DNA seen through the eyes of a coder - http://bit.ly/2DJbfcI

LEARNING Using Self-Organizing Maps to solve the Traveling Salesman Problem - the Traveling Salesman Problem is a well known challenge in Computer Science: it consists on finding the shortest route possible that traverses all cities in a given map only once. Although its simple explanation, this problem is, indeed, NP-Complete. This implies that the difficulty to solve it increases rapidly with the number of cities, and we do not know in fact a general solution that solves the problem. For that reason, we currently consider that any method able to find a sub-optimal solution is generally good enough (we cannot verify if the solution returned is the optimal one most of the times). http://bit.ly/2DB4aqx

Speedy Neural Networks for Smart Auto-Cropping of Images - The ability to share photos directly on Twitter has existed since 2011 and is now an integral part of the Twitter experience. Today, millions of images are uploaded to Twitter every day. However, they can come in all sorts of shapes and sizes, which presents a challenge for rendering a consistent UI experience. The photos in your timeline are cropped to improve consistency and to allow you to see more Tweets at a glance. How do we decide what to crop, that is, which part of the image do we show you? http://bit.ly/2Bv77XU

How to solve 90% of NLP problems: a step-by-step guide - Using Machine Learning to understand and leverage text. http://bit.ly/2neSz9I Faster R-CNN: Down the rabbit hole of modern object detection http://bit.ly/2DMW7uq

Deep Learning - learn how to use Keras and Tensorflow to apply deep learning to computer vision problems. Great set of intro videos + exercises by Kaggle Learn. http://bit.ly/2Bw469J

RESOURCES Data Science at the Command Line - free ebook http://bit.ly/2DFUKyg

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 19 Jan 2018

Hi! This week we focused on learning resources. A lot of great videos from ICML 2017 are already published, also Kaggle prepared cool interactive tutorials waiting for AI developers. Enjoy your weekend reading other AI news and don’t forget to share it with your friends ;)

GENERAL Cloud AutoML: Making AI accessible to every business - Google's Cloud Automl Vision system -- a machine-learning-based image classifier -- is now available to the general public; anyone can sign up to the program, upload a set of 20-10,000 images and train a new model with them, which they can then use. http://bit.ly/2Dt1ob6

Alibaba neural network defeats human in global reading test - Chinese tech giant's research unit says its deep neural network model is the first to beat humans in the Stanford Question Answering Dataset, but is listed first alongside Microsoft on the latest rankings. http://zd.net/2ETgkLR

Stanford's AI Predicts Death for Better End-of-Life Care - Using artificial intelligence to predict when patients may die sounds like an episode from the dystopian science fiction TV series “Black Mirror.” But Stanford University researchers see this use of AI as a benign opportunity to help prompt physicians and patients to have necessary end-of-life conversations earlier. http://bit.ly/2mR1SfN

Slack Hopes Its AI Will Keep You from Hating Slack - The fastest-growing business app is relying on machine-learning tricks to fend off a deluge of messages—as well as competition from Facebook and Microsoft. http://bit.ly/2DsVYfd

PROGRAMMING A Faster Pytorch Implementation of Faster R-CNN - This project is a faster pytorch implementation of faster R-CNN, aimed to accelerating the training of faster R-CNN object detection models. http://bit.ly/2DsWeeb

LEARNING Official Kaggle Learn - Interactive tutorials on Machine Learning, Deep Learning, R, and Data Visualization. http://bit.ly/2mV2xxY

VIDEOS ICML | 2017 Thirty-fourth International Conference on Machine Learning - all videos http://bit.ly/2rl85q9

MIT 6.S094 Deep Learning for Self-Driving Cars 2018 (ep. 1) - This class is free and open to everyone. It is an introduction to the practice of deep learning through the applied theme of building a self-driving car. http://bit.ly/2Du5qPC

PAPERS Game-theory insights into asymmetric multi-agent games - how two intelligent systems behave and respond in a particular type of situation known as an asymmetric game, which include Leduc poker and various board games such as Scotland Yard. Asymmetric games also naturally model certain real-world scenarios such as automated auctions where buyers and sellers operate with different motivations. http://bit.ly/2FWqfl5

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 12 Jan 2018

Hi! Looks like we are back on track with AI News. This week you can read about turning design mockups directly into code and how Google implemented continous low-power music recognition in Pixel 2. If you have more time, i strongly suggest watching presentation about turning Integrated Circuits into logic schemes. For Silicon Valley fans there is a CVPR 2018 Learned Image Compression Challenge, where you can show your middle-out algorithm and maybe even beat Pied Piper ;) Enjoy your weekend reading other AI news and don’t forget to share it with your friends 😉

GENERAL

Japanese scientists just used AI to read minds and it’s amazing — Imagine a reality where computers can visualize what you are thinking.

Sound far out? It’s now closer to becoming a reality thanks to four scientists at Kyoto University in Kyoto, Japan. In late December, Guohua Shen, Tomoyasu Horikawa, Kei Majima and Yukiyasu Kamitani releasedthe results of their recent research on using artificial intelligence to decode thoughts on the scientific platform, BioRxiv. http://cnb.cx/2D91Lra

Laundroid — a $16,000 robot that uses artificial intelligence to sort and fold laundry. http://bit.ly/2r2Nyq8

Neural Style Transfer for Musical Melodies is hard — int this great post, authors describe why style transfer in music is not as simple as in case of pictures. http://bit.ly/2D817dC

The Google Brain Team — Looking Back on 2017 (Part 1 & 2) http://bit.ly/2mB2xmD http://bit.ly/2FuIcXN

PROGRAMMING

Turning Design Mockups Into Code With Deep Learning — Within three years deep learning will change front-end development. It will increase prototyping speed and lower the barrier for building software. The field took off last year when Tony Beltramelli introduced the pix2code paper and Airbnb launched sketch2code. Currently, the largest barrier to automating front-end development is computing power. However, we can use current deep learning algorithms, along with synthesized training data, to start exploring artificial front-end automation right now. http://bit.ly/2qYHMFV

TensorFlow World Resources — Organized & Useful Resources about Deep Learning with TensorFlow http://bit.ly/2FuUMpX

Tricking Neural Networks: Create your own Adversarial Examples — Assassination by neural network. Sound crazy? Well, it might happen someday, and not in the way you may think. Of course neural networks could be trained to pilot drones or operate other weapons of mass destruction, but even an innocuous (and presently available) network trained to drive a car could be turned to act against its owner. This is because neural networks are extremely susceptible to something called adversarial examples. http://bit.ly/2D8q73y

VIDEOS

Reading Silicon: How to Reverse Engineer Integrated Circuits — Ken Shirriff has seen the insides of more integrated circuits than most people have seen bellybuttons. (This is an exaggeration.) But the point is, where we see a crazy jumble of circuitry, Ken sees a riddle to be solved, and he’s got a method that guides him through the madness. http://bit.ly/2FuYuzZ

PAPERS

Now Playing: Continuous low-power music recognition — Existing music recognition applications require both user activation and a connection to a server that performs the actual recognition. In this paper authors present a low power music recognizer that runs entirely on a mobile phone and automatically recognizes music without requiring any user activation. A small music detector runs continuously on the mobile phone’s DSP (digital signal processor) chip and only wakes main the processor when it is confident that music is present. http://bit.ly/2mxeeKQ

Learning Tree-based Deep Model for Recommender Systems — authors propose a novel recommendation method based on tree. With user behavior data, the tree based model can capture user interests from coarse to fine, by traversing nodes top down and make decisions whether to pick up each node to user. Compared to traditional model-based methods like matrix factorization (MF), tree based model does not have to fetch and estimate each item in the entire set. Instead, candidates are drawn from subsets corresponding to user’s high-level interests, which is defined by the tree structure. http://bit.ly/2qZrigM

Adversarial Spheres — State of the art computer vision models have been shown to be vulnerable to small adversarial perturbations of the input. In other words, most images in the data distribution are both correctly classified by the model and are very close to a visually similar misclassified image. Despite substantial research interest, the cause of the phenomenon is still poorly understood and remains unsolved. Authors hypothesize that this counter intuitive behavior is a naturally occurring result of the high dimensional geometry of the data manifold. http://bit.ly/2EB9Gtu

CHALLENGES

Introducing the CVPR 2018 Learned Image Compression Challenge — To encourage progress in this field, Google, in collaboration with ETH and Twitter, is sponsoring the Workshop and Challenge on Learned Image Compression (CLIC) at the upcoming 2018 Computer Vision and Pattern Recognition conference (CVPR 2018). Training data is already available on that site. The test set will be released on February 15 and the deadline for submitting the compressed versions of the test set is February 22. http://bit.ly/2D8DTTu

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 5 Jan 2018

Welcome in 2018 and a Happy New Year! New AI weekly focuses mainly on China (probably biggest investor in AI right now) and some predictions for upcoming months. But that’s not all, you can find here also new cool summaries of AI related events and libraries from 2017. Enjoy your weekend reading other AI news and don’t forget to share it with your friends 😉

GENERAL

The 5 Most Interesting Artificial Intelligence Trends for Entrepreneurs to Follow in 2018 — The current pace of innovation makes it almost impossible to stay on top of the AI trends, but understanding these advancements is a must for business owners who want to stay ahead. http://bit.ly/2lYtob4

China is building a giant $2.1 billion research park dedicated to developing A.I. — China is planning to build a 13.8 billion yuan ($2.1 billion) technology park dedicated to developing artificial intelligence (AI), state-backed news agency Xinhua reported Wednesday. The campus will be constructed within five years and situated in the suburban Mentougou district in western Beijing. It will cover 54.87 hectares, Xinhua said. The technology park will be home to around 400 businesses and is expected to create an annual output value of about 50 billion yuan. http://cnb.cx/2EbXe3p

These are the two books that are helping Xi Jinping understand AI — China’s president Xi Jinping is an avid reader. He peppers his speeches with quotes from his favorite writers, including Charles Dickens, Victor Hugo, and Paulo Coelho. His annual New Year’s Day greetings offers a rare look into the secrets of his office bookshelf — something Chinese netizens enthusiastically examine each January. http://bit.ly/2CLqpOv

How an A.I. ‘Cat-and-Mouse Game’ Generates Believable Fake Photos — At a lab in Finland, a small team of Nvidia researchers recently built a system that can analyze thousands of (real) celebrity snapshots, recognize common patterns, and create new images that look much the same — but are still a little different. The system can also generate realistic images of horses, buses, bicycles, plants and many other common objects. http://nyti.ms/2Eb1Qqc

AI in 2018: Google seeks to turn early focus on AI into cash — Alphabet has spent billions injecting machine learning into all aspects of its business, though returns will be hard to track http://on.mktw.net/2F2jduM

How Do You Vote? 50 Million Google Images Give a Clue — What vehicle is most strongly associated with Republican voting districts? Extended-cab pickup trucks. For Democratic districts? Sedans. Those conclusions may not be particularly surprising. After all, market researchers and political analysts have studied such things for decades. But what is surprising is how researchers working on an ambitious project based at Stanford University reached those conclusions: by analyzing 50 million images and location data from Google Street View, the street-scene feature of the online giant’s mapping service. http://nyti.ms/2CJilgU

Deep learning sharpens views of cells and genes — Eyes are said to be the window to the soul — but researchers at Google see them as indicators of a person’s health. The technology giant is using deep learning to predict a person’s blood pressure, age and smoking status by analysing a photograph of their retina. Google’s computers glean clues from the arrangement of blood vessels — and a preliminary study suggests that the machines can use this information to predict whether someone is at risk of an impending heart attack. http://go.nature.com/2D0CNqN

Statistical Computing for Scientists and Engineers — full course with videos, slides and homework http://bit.ly/2lXcl9e

AI and Deep Learning in 2017 — A Year in Review by Denny Britz http://bit.ly/2CWbwWa

PROGRAMMING

30 Amazing Machine Learning Projects for the Past Year (v.2018) — For the past year, we’ve compared nearly 8,800 open source Machine Learning projects to pick Top 30 (0.3% chance). This is an extremely competitive list and it carefully picks the best open source Machine Learning libraries, datasets and apps published between January and December 2017. Mybridge AI evaluates the quality by considering popularity, engagement and recency. To give you an idea about the quality, the average number of Github stars is 3,558. http://bit.ly/2AxQ2fo

Improving your data science workflow with Docker — Containerization is a trend that is taking the tech world by storm, but how can you, a data scientist, use it to improve your workflow? Let’s start with some basics of containerization and specifically Docker and then we’ll look at a couple of use cases for containerized docker science. http://bit.ly/2EbbadF

TensorFlow 1.5.0 Release Candidate — http://bit.ly/2lYBYGA

AIGaming — a platform that allows computer programs — also known as bots, to play each other at challenging games to win bitcoin. http://bit.ly/2m1TgTw

wav2letter — Facebook AI Research Automatic Speech Recognition Toolkit http://bit.ly/2m06Vua

PAPERS

DeepMind Control Suite — The DeepMind Control Suite is a set of continuous control tasks with a standardised structure and interpretable rewards, intended to serve as performance benchmarks for reinforcement learning agents. The tasks are written in Python and powered by the MuJoCo physics engine, making them easy to use and modify. http://bit.ly/2AydWaJ

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh

[tomaszbaranowicz-blog]

AI Weekly 29 Dec 2017

Hi, last week of 2017 and last AI weekly (this year) goes straight to you. This week brought us not many general news in AI industry, but hopefully cool videos and papers will satisfy you. Enjoy your weekend reading AI news and don’t forget to share it with your friends 😉

GENERAL

Top 3 Technology Trends For 2018, Which Will Be A Game Changer ! http://bit.ly/2pYVnfI

The AI chip startup explosion is already here — All eyes may have been on Nvidia this year as its stock exploded higher thanks to an enormous amount of demand across all fronts: gaming, an increased interest in data centers, and its major potential applications in AI. But while Nvidia’s stock price and that chart may have been one of the more eye-popping parts of 2017, a year when AI continued its march toward being omnipresent in technology, something a little more subtle was happening in the AI world that may have even deeper ramifications. https://yhoo.it/2zNxRSy

PROGRAMMING

STYLE2PAINTS 2.0 — The AI which can paint on a sketch according to a given specific color style. http://bit.ly/2DwsTMl

How to train a Deep Neural Network using only TensorFlow C++ — The core of TensorFlow (TF) is built using C++, yet lots of conveniences are only available in the python API. This blog post goal was to implement the same basic Deep Neural Network (DNN) using only the TF C++ API and then using only CuDNN. http://bit.ly/2pUhlR4

VIDEOS

Superhuman AI for heads-up no-limit poker: Libratus beats top professionals — This talk gives a high-level explanation of Libratus, the first AI to defeat top humans in no-limit poker. A paper on the AI was published in Science in 2017. http://bit.ly/2CjnfQL

JavaDay UA 2017: Machine Learning and AI in Java on AWS (Julien Simon) http://bit.ly/2Ca929B

PAPERS

Visualizing the Loss Landscape of Neural Nets — Neural network training relies on our ability to find “good” minimizers of highly non-convex loss functions. It is well known that certain network architecture designs (e.g., skip connections) produce loss functions that train easier, and well-chosen training parameters (batch size, learning rate, optimizer) produce minimizers that generalize better. However, the reasons for these differences, and their effect on the underlying loss landscape, is not well understood. In this paper, authors explore the structure of neural loss functions, and the effect of loss landscapes on generalization, using a range of visualization methods. http://bit.ly/2ll4oLx

Advances in Pre-Training Distributed Word Representations — Many Natural Language Processing applications nowadays rely on pre-trained word representations estimated from large text corpora such as news collections, Wikipedia and Web Crawl. In this paper, authors show how to train high-quality word vector representations by using a combination of known tricks that are however rarely used together. http://bit.ly/2lkySNA

If you want to be always on time with AI weekly, feel free to follow it on fb http://bit.ly/2yGcZhh