Learning Problem-agnostic Speech Representations from Multiple Self-supervised Tasks
Posted on06/04/2019
Learning good representations without supervision is still an open issue in machine learning, and is particularly challenging for speech signals, which are... Read More
InfoMask: Masked Variational Latent Representation to Localize Chest Disease
Posted on28/03/2019
The scarcity of richly annotated medical images is limiting supervised deep learning based solutions to medical image analysis tasks, such as localizing... Read More
INFER: INtermediate representations for FuturE pRediction
Posted on26/03/2019
In urban driving scenarios, forecasting future trajectories of surrounding vehicles is of paramount importance. While several approaches for the problem have been... Read More
Unreproducible Research is Reproducible
Posted on02/03/2019
The apparent contradiction in the title is a wordplay on the different meanings attributed to the word reproducible across different scientific fields.... Read More
Revisiting Reweighted Wake-Sleep for Models with Stochastic Control Flow
Posted on01/03/2019
Stochastic control-flow models (SCFMs) are a class of generative models that involve branch- ing on choices from discrete random vari- ables. Amortized... Read More
The Thermodynamic Variational Objective
Posted on01/03/2019
We introduce the thermodynamic variational objective (TVO) for learning in both continuous and discrete deep generative models. The TVO arises from a... Read More
CoQA: A Conversational Question Answering Challenge
Posted on01/03/2019
The goal of the CoQA (Conversational Question Answering) challenge is to measure the ability of machines to understand a text passage and... Read More
Hyperbolic Discounting and Learning over Multiple Horizons
Posted on19/02/2019
Reinforcement learning (RL) typically defines a discount factor as part of the Markov Decision Process. The discount factor values future rewards by... Read More
Separating value functions across time-scales
Posted on05/02/2019
In many finite horizon episodic reinforcement learning (RL) settings, it is desirable to optimize for the undiscounted return – in settings like... Read More
Separable value functions across time-scales
Posted on05/02/2019
In many finite horizon episodic reinforcement learning (RL) settings, it is desirable to optimize for the undiscounted return – in settings like... Read More