This code implements Prioritized Level Replay, a method for sampling training levels for reinforcement learning agents that exploits the fact that not all levels are equally useful for agents to learn from during training.
Code associated with paper "High-Dimensional Contextual Policy Search with Unknown Context Rewards using Bayesian Optimization"
Code supporting the paper Adversarial Attacks on Contextual Bandits accepted at Neurips 2020.
We are implementing differentiable models of robot manipulators, which allows us to learn typically assumed to be known models of robots for control and motion planning.
Code for ICDM 2020 paper Context-aware Deep Representation Learning for Geo-spatiotemporal Analysis
I have implemented in both python and R two papers for estimating subgroup means under misclassification, which are useful for data analyses. T. K. MAK, W. K. LI, A new method for estimating subgroup means under misclassification, Biometrika, Volume 75, Issue 1, March 1988, Pages 105–111, https//doi.org/10.1093/biomet/75.1.105 Selén, Jan. “Adjusting for Errors in Classification and Measurement in the Analysis of Partly and Purely Categorical Data.” Journal of the American Statistical Association, vol. 81, no. 393, 1986, pp. 75–81. JSTOR, www.jstor.org/stable/2287969. Accessed 10 Aug. 2020.
This repository contains a set of tools to render Binaural Room Impulse Responses (BRIR) using the Spatial Decomposition Method (SDM).The implementation features a series of improvements presented in Amengual et al. 2020, such as quantization of the direction of arrival (DOA) estimates to improve the spectral properties of the rendered BRIRs, or RTMod and RTMod+AP equalization for the late reverberation.The repository also contains the necessary files to 3D print an array holder of optimized topology for the estimation of DOA information.
A machine learning system that captures the dynamics of physical systems on the basis of observed trajectories via Hamiltionian modeling and symplectic integration.
A python library that provides common I/O interface across different storage backends.
Evolutionary Scale Modeling (esm): Pretrained language models for proteins
Code for paper <PointContrast: Unsupervised Pretraining for 3D Point Cloud Understanding>
Real Time Speech Enhancement in the Waveform Domain (Interspeech 2020)We provide a PyTorch implementation of the paper Real Time Speech Enhancement in the Waveform Domain. In which, we present a causal speech enhancement model working on the raw waveform that runs in real-time on a laptop CPU. The proposed model is based on an encoder-decoder architecture with skip-connections. It is optimized on both time and frequency domains, using multiple loss functions. Empirical evidence shows that it is capable of removing various kinds of background noise including stationary and non-stationary noises, as well as room reverb. Additionally, we suggest a set of data augmentation techniques applied directly on the raw waveform which further improve model performance and its generalization abilities.
Code to reproduce the results in the FAIR research papers "Semi-Supervised Learning of Visual Features by Non-Parametrically Predicting View Assignments with Support Samples" https://arxiv.org/abs/2104.13963 and "Supervision Accelerates Pre-training in Contrastive Semi-Supervised Learning of Visual Representations" https://arxiv.org/abs/2006.10803
Time-based Sequence Model for Personalization and Recommendation Systems
Deep image generation is becoming a tool to enhance artists and designers creativity potential. In this paper, we aim at making the generation process more structured and easier to interact with. Inspired by vector graphics systems, we propose a new deep image reconstruction paradigm where the outputs are composed from simple layers, defined by their color and a vector transparency mask.
