Microsoft at ICML 2019
Microsoft is excited to be a Gold sponsor of ICML. We will have over 100 Microsoft attendees present at the conference. Stop by our booth (#310) to chat with our experts, see demos of our…
Dead-ends and Secure Exploration in Reinforcement Learning [1.0]
This repository hosts the code for the following ICML 2019 paper: Dead-ends and Secure Exploration in Reinforcement Learning
Fast and Flexible Multi-Task Classification Using Conditional Neural Adaptive Processes
This talk will describe our recent work on designing image classification systems that, after an initial multi-task training phase, can automatically adapt to new tasks encountered at test time. I will introduce an approach that…
From automatic differentiation to message passing
Automatic differentiation is an elegant technique for converting a computable function expressed as a program into a derivative-computing program with similar time complexity. It does not execute the original program as a black-box, nor does…
Low Latency Privacy Preserving Inference
Reliability in Reinforcement Learning
Reinforcement Learning (RL), much like scaling a 3,000-foot rock face, is about learning to make sequential decisions. The list of potential RL applications is expansive, spanning robotics (drone control), dialogue systems (personal assistants, automated call…
A phonetic matching made inˈhɛvən
Recently, Microsoft Research Montréal open sourced a phonetic matching component used previously in Maluuba Inc.’s natural language understanding platform. The library contains string comparison utilities that operate on a phoneme level as opposed to a…
Sharing Updatable Models (SUM) on Blockchain
A framework to host and train publicly available machine learning models while crowdsourcing a dataset. Ideally, using a model for prediction is free. An incentive mechanism validates added data.