Zobrazeno 1 - 10
of 20
pro vyhledávání: '"Agrawal, Kumar Krishna"'
Non-contrastive self-supervised learning (NC-SSL) methods like BarlowTwins and VICReg have shown great promise for label-free representation learning in computer vision. Despite the apparent simplicity of these techniques, researchers must rely on se
Externí odkaz:
http://arxiv.org/abs/2312.10725
Autor:
Berlot-Attwell, Ian, Agrawal, Kumar Krishna, Carrell, A. Michael, Sharma, Yash, Saphra, Naomi
Although modern neural networks often generalize to new combinations of familiar concepts, the conditions that enable such compositionality have long been an open question. In this work, we study the systematicity gap in visual question answering: th
Externí odkaz:
http://arxiv.org/abs/2311.08695
Autor:
Sela, Gur-Eyal, Gog, Ionel, Wong, Justin, Agrawal, Kumar Krishna, Mo, Xiangxi, Kalra, Sukrit, Schafhalter, Peter, Leong, Eric, Wang, Xin, Balaji, Bharathan, Gonzalez, Joseph, Stoica, Ion
Efficient vision works maximize accuracy under a latency budget. These works evaluate accuracy offline, one image at a time. However, real-time vision applications like autonomous driving operate in streaming settings, where ground truth changes betw
Externí odkaz:
http://arxiv.org/abs/2208.07479
Representation learning that leverages large-scale labelled datasets, is central to recent progress in machine learning. Access to task relevant labels at scale is often scarce or expensive, motivating the need to learn from unlabelled datasets with
Externí odkaz:
http://arxiv.org/abs/2202.05808
We introduce the "inverse bandit" problem of estimating the rewards of a multi-armed bandit instance from observing the learning process of a low-regret demonstrator. Existing approaches to the related problem of inverse reinforcement learning assume
Externí odkaz:
http://arxiv.org/abs/2106.14866
While normalizing flows have led to significant advances in modeling high-dimensional continuous distributions, their applicability to discrete distributions remains unknown. In this paper, we show that flows can in fact be extended to discrete event
Externí odkaz:
http://arxiv.org/abs/1905.10347
Autor:
Engel, Jesse, Agrawal, Kumar Krishna, Chen, Shuo, Gulrajani, Ishaan, Donahue, Chris, Roberts, Adam
Efficient audio synthesis is an inherently difficult machine learning task, as human perception is sensitive to both global structure and fine-scale waveform coherence. Autoregressive models, such as WaveNet, model local structure at the expense of g
Externí odkaz:
http://arxiv.org/abs/1902.08710
Autor:
Kostrikov, Ilya, Agrawal, Kumar Krishna, Dwibedi, Debidatta, Levine, Sergey, Tompson, Jonathan
We identify two issues with the family of algorithms based on the Adversarial Imitation Learning framework. The first problem is implicit bias present in the reward functions used in these algorithms. While these biases might work well for some envir
Externí odkaz:
http://arxiv.org/abs/1809.02925
Deep reinforcement learning has led to several recent breakthroughs, though the learned policies are often based on black-box neural networks. This makes them difficult to interpret and to impose desired specification constraints during learning. We
Externí odkaz:
http://arxiv.org/abs/1807.00403
In this paper, we introduce Key-Value Memory Networks to a multimodal setting and a novel key-addressing mechanism to deal with sequence-to-sequence models. The proposed model naturally decomposes the problem of video captioning into vision and langu
Externí odkaz:
http://arxiv.org/abs/1611.06492