News in artificial intelligence and machine learning
…week of 12th October 2015. If you prefer this to land straight in your inbox, drop me your details here!
Technology news, trends and opinions
- How artificial intelligence is moving from the lab to your kid’s playroom, The Washington Post. Your daughter can now have a conversation with her Barbie!
- Six human skills Facebook’s artificial intelligence tech is slowly mastering, NY Business Journal
- Hackers can silently control Siri from 16 feet away, Wired
- Publicis.Sapient Acquires Minority Stake in Artificial Intelligence Company Lucid, The Wall Street Journal. Media behemoth looking to power their marketing and media buying tools.
- The Challenges Of Building AI Apps, by Mike Chalfen, TechCrunch
- The 7 Parts of Your Business That Need Automation Right Now, Inc.
- IBM opens Watson competency centre in Victoria, ZDNet
- Why Artificial Intelligence Should Read and Write Stories, Huffington Post
- Should cars be fully driverless? No, says an engineer and historian
- Intelligence amplification will make us as smart as the machines
- Researchers Digitise Part of a Rat Brain (paper here)
- Thought process: Building an artificial brain, Paul Allen’s $500 million quest to dissect the mind and code a new one from scratch
Research and development
- Narrative Science Systems: A Review, the authors discuss the challenges in building systems to generate long-form natural language and current approaches to this end.
- Layer-Specific Adaptive Learning Rates for Deep Networks, University of Maryland and US Naval Academy. The authors present an approach for overcoming the week to month-long training times for complex deep networks.
- Structured Memory for Neural Turing Machines, IBM Watson. The authors propose new architectures for the long short-term memory component of the DeepMind Neural Turing Machine to alleviate overfitting and increase accuracy.
- UAVs using Bayesian Optimization to Locate WiFi Devices, EPFL
- Evaluating Real-time Anomaly Detection Algorithms, Numenta provide a standard, open source framework to compare and evaluate different algorithms for detecting anomalies in streaming data.
- ADAAPT: A Deep Architecture for Adaptive Policy Transfer from Multiple Sources, IIT Madras. This work presents an approach for transfer learning that avoids negative transfer (which slows learning) and enables the transfer from multiple source tasks to specific parts of the target state space.
- Better Exploiting OS-CNNs for Better Event Recognition in Images, Shenzhen Institutes of Advanced Technology. This paper shows how one can use an object-focused convolutional neural network (CNN) and a scene-focused CNN to extract information to power cultural event recognition in images.
- Deep convolutional neural networks for pedestrian detection, Politecnico di Milano. Continuing from last weeks research for autonomous vehicles, this paper presents a CNN using both handcrafted and learned features.
Anything else catch your eye? Drop me a line on @nathanbenaich.
Playfair Capital is actively seeking out entrepreneurs building companies that change the way we live, work and play using deep learning, machine learning, artificial intelligence, computer vision, speech and NLP.