In the News
Deep Learning Machine teaches itself chess in 72 hours
In a world first, an artificial intelligence machine plays chess by evaluating the board rather than using brute force to work out every possible move.
Apple to increase Artificial Intelligence staff in challenge to Google
Apple has ramped up its hiring of artificial intelligence experts, recruiting from PhD programs, posting dozens of job listings and greatly increasing the size of its AI staff, a review of hiring sites suggests and numerous sources confirm.
Apple also unveiled new products such as the Apple TV, showing off both the promises and limitations of AI. Oh and Google's Eric Schmidt took a potshot at Apple Music
Facebook's problem: Its algorithms aren't smart enough
Facebook has a billion users in a single day. Close to a billion photos posted every day. Three billion videos viewed every day. And driving all of that is a powerful set of algorithms that determine what we see and when. But is that a good thing or a bad thing? It’s complicated.
Intelligent machines: Call for a ban on robots designed as sex toys
A campaign has been launched calling for a ban on the development of robots that can be used for sex. Such a use of the technology is unnecessary and undesirable, said campaign leader Dr Kathleen Richardson. Sex dolls already on the market are becoming more sophisticated and some are now hoping to build artificial intelligence into their products.
Also in the news this week...
- Tencent claims it is getting closer to automated journalism as its Robot reporter 'Dreamwriter' churns out good 1,000-word news stories in 60 seconds
- Toyota invests heavily in AI, announcing a joint Stanford-MIT research center for intelligence robotics and autonomous cars
- Hitachi announced that it is using AI to organize work in factories (people talk about the 1st AI boss)
- Paxata raises $18M for software that makes data analysis easier
- Deepomatic raises $1.4M to develop its Deep Learning visual search technology
Learning
26 things I learned in the Deep Learning Summer School
List of small nuggets of information and insights that will be relevant to anybody working in the Machine Learning / Neural Network fields. Interesting read.
Recurrent Neural Networks Tutorial, Part 1 – Introduction
Recurrent Neural Networks (RNNs) are popular models that have shown great promise in many NLP tasks. This is the beginning of a series of tutorials on RNN, starting with the implementation of a recurrent neural network based language model.
Modern Methods for Sentiment Analysis
Sentiment analysis is a common application of Natural Language Processing (NLP) methodologies, particularly classification, whose goal is to extract the emotional content in text. This article introduces techniques such as Word2Vec and Doc2Vec.
Deep Style: Inferring the unknown to predict Fashion
One path to better recommendations involves creating an automated process to understand and quantify the style our inventory and clients at a fundamental level. Few would doubt that fashion is primarily a visual art form, so in order to achieve this goal we must first develop a way to interpret the style within images of clothing. In this post we'll look specifically at how to build an automated process using photographs of clothing to quantify the style of some of items in our collection.
Letting users choose Recommender Algorithms - an Experimental Study
We gave users the ability to change the algorithm providing their movie recommendations and studied how they make use of this power. A substantial portion of the user base (25%) used the recommender-switching feature. The majority of users who used the control only switched algorithms a few times, trying a few out and settling down on an algorithm that they would leave alone.
Inferring Algorithmic Patterns with Stack
We show that certain simple sequential patterns cannot be learned by popular deep learning approaches. Standard artificial neural networks cannot learn simple fundamental concepts such as memorization of sequence of symbols as they rely mostly on remembering previously seen patterns frequently appearing in the training data. We propose a novel sequence prediction approach which has the capability to learn these simple concepts.
Software tools & code
The power of Spatial Transformer Networks
Spatial Transformer Networks could be the future of Convolutional Neural Networks. Here, a team at Moodstock explores roadsign classification with Transformer Networks on top of standard ConvNets.
Analyzing 1.7 billion Reddit comments with Blaze and Impala
Learn how to interactively query and explore a data set of approximately 1.7 billion comments (975 GB uncompressed) from Reddit. This posts shows how S3 and Hadoop on Amazon EC2 were used along with Impala and Blaze to perform the analysis.
Deep Hear - Composing and harmonizing music with neural networks
Post showing how a network was trained to generate random bars of music, based on Scott Joplin's ragtime music. It is a fully connected Deep Belief Network, set up to perform an auto-encoding task.
Brains & Neurons
Towards Biologically Plausible Deep Learning
Neuroscientists have long criticised deep learning algorithms as incompatible with current knowledge of neurobiology. This article explores more biologically plausible versions of deep representation learning, focusing mostly on unsupervised learning.
Some thoughts
Reanimation of the tea party & riddle scene from Alice in Wonderland
Short video showing a famous scene of Alice in Wonderland, re-styled to look like 17 famous paintings (Pablo Picasso, Georgia O'Keeffe, S.H. Raza , Hokusai, Frida Kahlo, Vincent van Gogh, Tarsila, Saloua Raouda Choucair, Lee Krasner, Sol Lewitt, Wu Guanzhong, Elaine de Kooning, Ibrahim el-Salahi, Minnie Pwerle, Jean-Michel Basquiat, Edvard Munch, Natalia Goncharova)
About
This newsletter is a weekly collection of AI news and resources.
If you find this newsletter worthwhile, please forward to your friends and colleagues, or share on your favorite network!
Share on Twitter · Share on Linkedin · Share on Google+
Suggestions or comments are more than welcome, just reply to this email. Thanks!