[edit]

# AISTATS 2017 Program of Events

## Best Paper Awards

A Sub-Quadratic Exact Medoid AlgorithmJames Newling, Francois Fleuret

Phase Retrieval Meets Statistical Learning Theory: A Flexible Convex Relaxation

Sohail Bahmani, Justin Romberg

Reparameterization Gradients through Acceptance- Rejection Sampling Algorithms

Christian Naesseth, Francisco Ruiz, Scott Linderman, David Blei

## 19-Apr (Wed)

**18:30-20:30**Registration Desk

## 20-Apr (Thu)

**7:30-8:50**Breakfast, Windows on the Green & Chart Room

**8-10**Registration Desk

**8:50-9pm**Welcome and award announcement

**9:00-10:00**Invited Talk: Csaba Szepesvari. Crystal Ballroom 1, 2

**Stochastic linear bandits.**See abstract. See slides.

**10:00-10:30**Coffee Break, Crystal Atrium

**10:30-12:10**

__Online Learning__, Crystal Ballroom 1, 2

*Session Chair: Csaba Szepesvari*

63 Linear Thompson Sampling Revisited

217 Horde of Bandits using Gaussian Markov Random Fields

225 The End of Optimism? An Asymptotic Analysis of Finite-Armed Linear Bandits

304 Improved Strongly Adaptive Online Learning using Coin Betting

**12:10-2:00**Lunch on your own

**1:00-3:00**Registration Desk

**2:00-3:40**

__Nonparametric methods__, Crystal Ballroom 1, 2

*Session Chair: Byron Boots*

97 Poisson intensity estimation with reproducing kernels

249 Attributing Hacks

248 Regression Uncertainty on the Grassmannian

401 Modal-set estimation with an application to clustering

**3:40-4:10**Coffee break, Crystal Atrium

**4:10-7:00**Poster Session (with light snacks), Crystal Ballroom 3, 4

See poster list.

## 21-Apr (Fri)

**7:30-9:00**Breakfast, Windows on the Green & Chart Room

**8-10**Registration Desk

**9:00-10:00**Invited Talk, Cynthia Rudin, Crystal Ballroom 1, 2

**What Are We Afraid Of?: Computational Hardness vs the Holy Grail of Interpretability in Machine Learning.**See abstract. See slides.

**10:00-10:30**Coffee Break, Crystal Atrium

**10:30-12:10**

__Theory__, Crystal Ballroom 1, 2

*Session Chair: Sanjoy Dasgupta*

94 Phase Retrieval Meets Statistical Learning Theory: A Flexible Convex Relaxation

68 A Sub-Quadratic Exact Medoid Algorithm

456 On the Interpretability of Conditional Probability Estimates in the Agnostic Setting

209 Beta calibration: a well-founded and easily implemented improvement on logistic calibration for binary classifiers

**12:10-2:00**Lunch on your own

**1:00-3:00**Registration Desk

**2:00-3:40**

__Approximate Inference and MCMC__, Crystal Ballroom 1, 2

*Session Chair: Simon Lacoste-Julien*

51 Annular Augmentation Sampling

101 Removing Phase Transitions from Gibbs Measures

170 Reparameterization Gradients through Acceptance-Rejection Sampling Algorithms

174 Asymptotically exact inference in differentiable generative models

**3:40-4:10**Coffee Break, Crystal Atrium

**4:10-7:00**Poster Session (with light snacks), Crystal Ballroom 3, 4

See poster list.

**7:15-9:00**Dinner Buffet, Panorama Ballroom

## 22-Apr (Sat)

**7:30-9:00**Breakfast, Panorama Ballroom C, D & Terrace

**8-10**Registration Desk

**9:00-10:00**Invited Talk: Sanjoy Dasgupta. Panorama Ballroom A, B

**Towards a Theory of Interactive Learning.**See abstract. See slides.

**10:00-10:30**Coffee Break, Panorama Foyer

**10:30-12:10**

__Bayesian Methods__, Panorama Ballroom A, B

*Session Chair: Rebecca Steorts*

420 Signal-based Bayesian Seismic Monitoring

180 Fast Bayesian Optimization of Machine Learning Hyperparameters on Large Datasets

82 Near-optimal Bayesian Active Learning with Correlated and Noisy Tests

298 On the Hyperprior Choice for the Global Shrinkage Parameter in the Horseshoe Prior

**12:10-1:30**Lunch on your own

**(note shorter lunch)**

**1:30-3:10**

__Large-scale learning__, Panorama Ballroom A, B

*Session Chair: Pradeep Ravikumar*

417 Communication-Efficient Learning of Deep Networks from Decentralized Data

520 Automated Inference with Adaptive Batches

224 Adaptive ADMM with Spectral Penalty Parameter Selection

372 Identifying groups of strongly correlated variables through Smoothed Ordered Weighted L_1-norms

**3:10-3:40**Coffee Break Panorama Foyer

**3:40-5:20**

__Sketching__, Panorama Ballroom A, B

*Session Chair: Anastasios (Tasos) Kyrillidis*

384 Sketching Meets Random Projection in the Dual: A Provable Recovery Algorithm for Big and High-dimensional Data

399 Sketchy Decisions: Convex Low-Rank Matrix Optimization with Optimal Storage

192 Co-Occurring Directions Sketching for Approximate Matrix Multiply

117 Random Consensus Robust PCA