Research Area
Year Published

219 Results

June 16, 2019

Leveraging the Present to Anticipate the Future in Videos

CVPR Precognition Workshop

Anticipating actions before they are executed is crucial for a wide range of practical applications including autonomous driving and robotics. While most prior work in this area requires partial observation of executed actions, in the paper we focus on anticipating actions seconds before they start. Our proposed approach is the fusion of a purely anticipatory model with a complementary model constrained to reason about the present.

By: Antoine Miech, Ivan Laptev, Josef Sivic, Heng Wang, Lorenzo Torresani, Du Tran

June 16, 2019

On the Idiosyncrasies of the Mandarin Chinese Classifier System

North American Chapter of the Association for Computational Linguistics (NAACL)

While idiosyncrasies of the Chinese classifier system have been a richly studied topic among linguists (Adams and Conklin, 1973; Erbaugh, 1986; Lakoff, 1986), not much work has been done to quantify them with statistical methods. In this paper, we introduce an information-theoretic approach to measuring idiosyncrasy; we examine how much the uncertainty in Mandarin Chinese classifiers can be reduced by knowing semantic information about the nouns that the classifiers modify.

By: Shijia Liu, Hongyuan Mei, Adina Williams, Ryan Cotterell

June 13, 2019

Multi-modal Content Localization in Videos Using Weak Supervision

International Conference on Machine Learning (ICML)

Identifying the temporal segments in a video that contain content relevant to a category or task is a difficult but interesting problem. This has applications in fine-grained video indexing and retrieval. Part of the difficulty in this problem comes from the lack of supervision since large-scale annotation of localized segments containing the content of interest is very expensive. In this paper, we propose to use the category assigned to an entire video as weak supervision to our model.

By: Gourab Kundu, Prahal Arora, Ferdi Adeputra, Polina Kuznetsova, Daniel McKinnon, Michelle Cheung, Larry Anazia, Geoffrey Zweig

June 11, 2019

ELF OpenGo: An Analysis and Open Reimplementation of AlphaZero

International Conference on Machine Learning (ICML)

The AlphaGo, AlphaGo Zero, and AlphaZero series of algorithms are remarkable demonstrations of deep reinforcement learning’s capabilities, achieving superhuman performance in the complex game of Go with progressively increasing autonomy. However, many obstacles remain in the understanding of and usability of these promising approaches by the research community.

By: Yuandong Tian, Jerry Ma, Qucheng Gong, Shubho Sengupta, Zhuoyuan Chen, James Pinkerton, Larry Zitnick

June 11, 2019

Adversarial Inference for Multi-Sentence Video Description

Conference Computer Vision and Pattern Recognition (CVPR)

While significant progress has been made in the image captioning task, video description is still in its infancy due to the complex nature of video data. Generating multi-sentence descriptions for long videos is even more challenging. Among the main issues are the fluency and coherence of the generated descriptions, and their relevance to the video.

By: Jae Sung Park, Marcus Rohrbach, Trevor Darrell, Anna Rohrbach

June 10, 2019

GEOMetrics: Exploiting Geometric Structure for Graph-Encoded Objects

International Conference on Machine Learning (ICML)

Mesh models are a promising approach for encoding the structure of 3D objects. Current mesh reconstruction systems predict uniformly distributed vertex locations of a predetermined graph through a series of graph convolutions, leading to compromises with respect to performance or resolution. In this paper, we argue that the graph representation of geometric objects allows for additional structure, which should be leveraged for enhanced reconstruction.

By: Edward J. Smith, Scott Fujimoto, Adriana Romero, David Meger

June 10, 2019

Unreproducible Research is Reproducible

International Conference on Machine Learning (ICML)

The apparent contradiction in the title is a wordplay on the different meanings attributed to the word reproducible across different scientific fields. What we imply is that unreproducible findings can be built upon reproducible methods. Without denying the importance of facilitating the reproduction of methods, we deem important to reassert that reproduction of findings is a fundamental step of the scientific inquiry.

By: Xavier Bouthillier, Cesar Laurent, Pascal Vincent

June 10, 2019

Garbage In, Reward Out: Bootstrapping Exploration in Multi-Armed Bandits

International Conference on Machine Learning (ICML)

We propose a bandit algorithm that explores by randomizing its history of rewards. Specifically, it pulls the arm with the highest mean reward in a non-parametric bootstrap sample of its history with pseudo rewards. We design the pseudo rewards such that the bootstrap mean is optimistic with a sufficiently high probability. We call our algorithm Giro, which stands for garbage in, reward out.

By: Branislav Kveton, Csaba Szepesvári, Sharan Vaswani, Zheng Wen, Mohammad Ghavamzadeh, Tor Lattimore

June 10, 2019

Mixture Models for Diverse Machine Translation: Tricks of the Trade

International Conference on Machine Learning (ICML)

We develop an evaluation protocol to assess both quality and diversity of generations against multiple references, and provide an extensive empirical study of several mixture model variants. Our analysis shows that certain types of mixture models are more robust and offer the best trade-off between translation quality and diversity compared to variational models and diverse decoding approaches.

By: Tianxiao Shen, Myle Ott, Michael Auli, Marc'Aurelio Ranzato

June 10, 2019

Non-Monotonic Sequential Text Generation

International Conference on Machine Learning (ICML)

Standard sequential generation methods assume a pre-specified generation order, such as text generation methods which generate words from left to right. In this work, we propose a framework for training models of text generation that operate in non-monotonic orders; the model directly learns good orders, without any additional annotation.

By: Sean Welleck, Kianté Brantley, Hal Daumé III, Kyunghyun Cho