A
Antonio Torralba
Researcher at Massachusetts Institute of Technology
Publications - 437
Citations - 105763
Antonio Torralba is an academic researcher from Massachusetts Institute of Technology. The author has contributed to research in topics: Computer science & Object detection. The author has an hindex of 119, co-authored 388 publications receiving 84607 citations. Previous affiliations of Antonio Torralba include Vassar College & Nvidia.
Papers
More filters
Posted Content
Predicting Motivations of Actions by Leveraging Text
TL;DR: The problem of predicting why a person has performed an action in images is introduced and results suggest that transferring knowledge from language into vision can help machines understand why people in images might be performing an action.
Book ChapterDOI
Comparing the Interpretability of Deep Networks via Network Dissection.
TL;DR: This chapter introduces Network Dissection, a general framework to quantify the interpretability of the units inside a deep convolutional neural networks (CNNs), and compares the different vocabularies of interpretable units as concept detectors emerged from the networks trained to solve different supervised learning tasks.
Posted Content
The Hessian Penalty: A Weak Prior for Unsupervised Disentanglement.
TL;DR: The Hessian Penalty as discussed by the authors is a simple regularization term that encourages the Hessian of a generative model with respect to its input to be diagonal, which can be applied to a wide range of deep generators with just a few lines of code.
Journal ArticleDOI
Improving Factuality and Reasoning in Language Models through Multiagent Debate
TL;DR: This paper proposed a "Society of minds" approach to improve language responses where multiple language model instances propose and debate their individual responses and reasoning processes over multiple rounds to arrive at a common final answer.
Posted Content
Rewriting a Deep Generative Model
TL;DR: In this paper, the authors introduce a new problem setting: manipulation of specific rules encoded by a deep generative model, and propose a formulation in which the desired rule is changed by manipulating a layer of a deep network as a linear associative memory.