Author pages are created from data sourced from our academic publisher partnerships and public sources.

- Publications
- Influence

Deep Sets

- M. Zaheer, S. Kottur, Siamak Ravanbakhsh, B. Póczos, R. Salakhutdinov, Alex Smola
- Mathematics, Computer Science
- NIPS
- 10 March 2017

We study the problem of designing models for machine learning tasks defined on sets. In contrast to the traditional approach of operating on fixed dimensional vectors, we consider objective functions… Expand

Stochastic Variance Reduction for Nonconvex Optimization

- S. Reddi, A. Hefny, S. Sra, B. Póczos, Alex Smola
- Computer Science, Mathematics
- ICML
- 19 March 2016

We study nonconvex finite-sum problems and analyze stochastic variance reduced gradient (SVRG) methods for them. SVRG and related methods have recently surged into prominence for convex optimization… Expand

Gradient Descent Provably Optimizes Over-parameterized Neural Networks

One of the mysteries in the success of neural networks is randomly initialized first order methods like gradient descent can achieve zero training loss even though the objective function is… Expand

MMD GAN: Towards Deeper Understanding of Moment Matching Network

- C. Li, Wei-Cheng Chang, Yu Cheng, Yiming Yang, B. Póczos
- Mathematics, Computer Science
- NIPS
- 24 May 2017

Generative moment matching network (GMMN) is a deep generative model that differs from Generative Adversarial Network (GAN) by replacing the discriminator in GAN with a two-sample test based on… Expand

Proximal Stochastic Methods for Nonsmooth Nonconvex Finite-Sum Optimization

- S. Reddi, S. Sra, B. Póczos, Alex Smola
- Mathematics, Computer Science
- NIPS
- 5 December 2016

We analyze stochastic algorithms for optimizing nonconvex, nonsmooth finite-sum problems, where the nonsmooth part is convex. Surprisingly, unlike the smooth case, our knowledge of this fundamental… Expand

High Dimensional Bayesian Optimisation and Bandits via Additive Models

- Kirthevasan Kandasamy, J. Schneider, B. Póczos
- Mathematics, Computer Science
- ICML
- 5 March 2015

Bayesian Optimisation (BO) is a technique used in optimising a $D$-dimensional function which is typically expensive to evaluate. While there have been many successes for BO in low dimensions,… Expand

Gradient Descent Learns One-hidden-layer CNN: Don't be Afraid of Spurious Local Minima

- S. Du, J. Lee, Yuandong Tian, B. Póczos, Amrendra K Singh
- Mathematics, Computer Science
- ICML
- 3 December 2017

We consider the problem of learning a one-hidden-layer neural network with non-overlapping convolutional layer and ReLU activation function, i.e., $f(\mathbf{Z}; \mathbf{w}, \mathbf{a}) = \sum_j… Expand

Neural Architecture Search with Bayesian Optimisation and Optimal Transport

- Kirthevasan Kandasamy, W. Neiswanger, Jeff Schneider, B. Póczos, E. Xing
- Computer Science, Mathematics
- NeurIPS
- 11 February 2018

Bayesian Optimisation (BO) refers to a class of methods for global optimisation of a function $f$ which is only accessible via point evaluations. It is typically used in settings where $f$ is… Expand

On Variance Reduction in Stochastic Gradient Descent and its Asynchronous Variants

- S. Reddi, A. Hefny, S. Sra, B. Póczos, Alex Smola
- Computer Science, Mathematics
- NIPS
- 23 June 2015

We study optimization algorithms based on variance reduction for stochastic gradient descent (SGD). Remarkable recent progress has been made in this direction through development of algorithms like… Expand

Variance Reduction in Stochastic Gradient Langevin Dynamics

- Kumar Avinava Dubey, S. Reddi, Sinead Williamson, B. Póczos, Alex Smola, E. Xing
- Computer Science, Medicine
- NIPS
- 1 December 2016

Stochastic gradient-based Monte Carlo methods such as stochastic gradient Langevin dynamics are useful tools for posterior inference on large scale datasets in many machine learning applications.… Expand