Filter by Research Area
Filter by Research Area
Year Published

236 Results

July 10, 2018

Modeling Others using Oneself in Multi-Agent Reinforcement Learning

International Conference on Machine Learning (ICML)

We consider the multi-agent reinforcement learning setting with imperfect information in which each agent is trying to maximize its own utility.

By: Roberta Raileanu, Emily Denton, Arthur Szlam, Rob Fergus
July 9, 2018

Mixed batches and symmetric discriminators for GAN training

International Conference on Machine Learning (ICML)

We propose to feed the discriminator with mixed batches of true and fake samples, and train it to predict the ratio of true samples in the batch.

By: Thomas Lucas, Corentin Tallec, Jakob Verbeek, Yann Ollivier
July 9, 2018

Composable Planning with Attributes

International Conference on Machine Learning (ICML)

We propose a method that learns a policy for transitioning between “nearby” sets of attributes, and maintains a graph of possible transitions.

By: Amy Zhang, Adam Lerer, Sainbayar Sukhbaatar, Rob Fergus, Arthur Szlam
July 9, 2018

Continuous Reasoning: Scaling the Impact of Formal Methods

Logic in Computer Science

This paper describes work in continuous reasoning, where formal reasoning about a (changing) codebase is done in a fashion which mirrors the iterative, continuous model of software development that is increasingly practiced in industry. We suggest that advances in continuous reasoning will allow formal reasoning to scale to more programs, and more programmers.

By: Peter O'Hearn
July 7, 2018

Algorithmic Regularization in Over-parameterized Matrix Sensing and Neural Networks with Quadratic Activations


We show that the gradient descent algorithm provides an implicit regularization effect in the learning of over-parameterized matrix factorization models and one-hidden-layer neural networks with quadratic activations.

By: Yuanzhi Li, Tengyu Ma, Hongyang Zhang
June 29, 2018

Understanding the Loss Surface of Neural Networks for Binary Classification

International Conference on Machine Learning (ICML)

Here we focus on the training performance of neural networks for binary classification, and provide conditions under which the training error is zero at all local minima of appropriately chosen surrogate loss functions.

By: Shiyu Liang, Ruoyu Sun, Yixuan Li, R. Srikant
June 19, 2018

LAMV: Learning to align and match videos with kernelized temporal layers

Computer Vision and Pattern Recognition (CVPR)

This paper considers a learnable approach for comparing and aligning videos. Our architecture builds upon and revisits temporal match kernels within neural networks: we propose a new temporal layer that finds temporal alignments by maximizing the scores between two sequences of vectors, according to a time-sensitive similarity metric parametrized in the Fourier domain.

By: Lorenzo Baraldi, Matthijs Douze, Rita Cucchiara, Hervé Jégou
June 19, 2018

Link and code: Fast indexing with graphs and compact regression codes

Computer Vision and Pattern Recognition (CVPR)

Similarity search approaches based on graph walks have recently attained outstanding speed-accuracy trade-offs, taking aside the memory requirements. In this paper, we revisit these approaches by considering, additionally, the memory constraint required to index billions of images on a single server.

By: Matthijs Douze, Alexandre Sablayrolles, Hervé Jégou
June 19, 2018

A Generative Adversarial Approach for Zero-Shot Learning from Noisy Texts

Computer Vision and Pattern Recognition (CVPR)

Most existing zero-shot learning methods consider the problem as a visual semantic embedding one. Given the demonstrated capability of Generative Adversarial Networks (GANs) to generate images, we instead leverage GANs to imagine unseen categories from text descriptions and hence recognize novel classes with no examples being seen.

By: Yizhe Zhu, Mohamed Elhoseiny, Bingchen Liu, Xi Peng, Ahmed Elgammal
June 18, 2018

DensePose: Dense Human Pose Estimation In The Wild

Computer Vision and Pattern Recognition (CVPR)

In this work we establish dense correspondences between an RGB image and a surface-based representation of the human body, a task we refer to as dense human pose estimation. We gather dense correspondences for 50K persons appearing in the COCO dataset by introducing an efficient annotation pipeline. We then use our dataset to train CNN-based systems that deliver dense correspondence ‘in the wild’, namely in the presence of background, occlusions and scale variations.

By: Riza Alp Guler, Natalia Neverova, Iasonas Kokkinos