1 code implementation • 28 Dec 2023 • Tim G. J. Rudner, Sanyam Kapoor, Shikai Qiu, Andrew Gordon Wilson
In this work, we approach regularization in neural networks from a probabilistic perspective and show that by viewing parameter-space regularization as specifying an empirical prior distribution over the model parameters, we can derive a probabilistically well-motivated regularization technique that allows explicitly encoding information about desired predictive functions into neural network training.
1 code implementation • NeurIPS 2023 • Shikai Qiu, Tim G. J. Rudner, Sanyam Kapoor, Andrew Gordon Wilson
Moreover, the most likely parameters under the parameter posterior do not generally correspond to the most likely function induced by the parameter posterior.
1 code implementation • 24 Nov 2022 • Sanae Lotfi, Marc Finzi, Sanyam Kapoor, Andres Potapczynski, Micah Goldblum, Andrew Gordon Wilson
While there has been progress in developing non-vacuous generalization bounds for deep neural networks, these bounds tend to be uninformative about why deep learning works.
1 code implementation • 20 May 2022 • Ravid Shwartz-Ziv, Micah Goldblum, Hossein Souri, Sanyam Kapoor, Chen Zhu, Yann Lecun, Andrew Gordon Wilson
Deep learning is increasingly moving towards a transfer learning paradigm whereby large foundation models are fine-tuned on downstream tasks, starting from an initialization learned on the source task.
1 code implementation • 30 Mar 2022 • Sanyam Kapoor, Wesley J. Maddox, Pavel Izmailov, Andrew Gordon Wilson
In Bayesian regression, we often use a Gaussian observation model, where we control the level of aleatoric uncertainty with a noise variance parameter.
1 code implementation • 31 Dec 2021 • Wesley J. Maddox, Sanyam Kapoor, Andrew Gordon Wilson
While recent work on conjugate gradient methods and Lanczos decompositions have achieved scalable Gaussian process inference with highly accurate point predictions, in several implementations these iterative methods appear to struggle with numerical instabilities in learning kernel hyperparameters, and poor test likelihoods.
no code implementations • 12 Nov 2021 • Sanyam Kapoor, Valerio Perrone
XGBoost, a scalable tree boosting algorithm, has proven effective for many prediction tasks of practical interest, especially using tabular datasets.
1 code implementation • 12 Jun 2021 • Sanyam Kapoor, Marc Finzi, Ke Alexander Wang, Andrew Gordon Wilson
State-of-the-art methods for scalable Gaussian processes use iterative algorithms, requiring fast matrix vector multiplies (MVMs) with the covariance kernel.
1 code implementation • 9 Jun 2020 • Sanyam Kapoor, Theofanis Karaletsos, Thang D. Bui
Through sequential construction of posteriors on observing data online, Bayes' theorem provides a natural framework for continual learning.
1 code implementation • 18 Oct 2019 • Ted Moskovitz, Rui Wang, Janice Lan, Sanyam Kapoor, Thomas Miconi, Jason Yosinski, Aditya Rawal
Standard gradient descent methods are susceptible to a range of issues that can impede training, such as high correlations and different scaling in parameter space. These difficulties can be addressed by second-order approaches that apply a pre-conditioning matrix to the gradient to improve convergence.
no code implementations • ICLR 2019 • Cinjon Resnick, Roberta Raileanu, Sanyam Kapoor, Alexander Peysakhovich, Kyunghyun Cho, Joan Bruna
Our contributions are that we analytically characterize the types of environments where Backplay can improve training speed, demonstrate the effectiveness of Backplay both in large grid worlds and a complex four player zero-sum game (Pommerman), and show that Backplay compares favorably to other competitive methods known to improve sample efficiency.
1 code implementation • 25 Jul 2018 • Sanyam Kapoor
Reinforcement Learning (RL) is a learning paradigm concerned with learning to control a system so as to maximize an objective over the long term.
Multi-agent Reinforcement Learning reinforcement-learning +1
1 code implementation • 18 Jul 2018 • Cinjon Resnick, Roberta Raileanu, Sanyam Kapoor, Alexander Peysakhovich, Kyunghyun Cho, Joan Bruna
Our contributions are that we analytically characterize the types of environments where Backplay can improve training speed, demonstrate the effectiveness of Backplay both in large grid worlds and a complex four player zero-sum game (Pommerman), and show that Backplay compares favorably to other competitive methods known to improve sample efficiency.