The NIPS Workshop on Multi-class and Multi-label Learning in Extremely Large Label Spaces
Friday, 9th December 2016, Barcelona, Spain
Extreme Delights
The workshop's videos are now available on YouTube. You can watch them by clicking on the talk title.

The workshop venue is Room 111.

Introduction

09:00 - 09:05 Samy Bengio (Google) Opening remarks
09:05 - 09:35 Thorsten Joachims (Cornell)Label Ranking with Biased Partial Feedback

Extreme Models with Linear Prediction Costs

09:35 - 09:50 Maximilian Alber (TU Berlin) Distributed Optimization of Multi-Class SVMs
09:50 - 10:05 Rohit Babbar (MPI) DiSMEC - Distributed Sparse Machines for Extreme Multi-label Classification
10:05 - 10:35 Inderjit Dhillon (UT Austin) A Primal and Dual Sparse Approach to Extreme Classification
 
10:35 - 11:00 Coffee break

Extreme Models with Sub-linear Prediction Costs

11:00 - 11:30 Manik Varma (Microsoft Research) Extreme Multi-label Loss Functions for Tagging, Ranking & Recommendation
11:30 - 11:45 Kalina Jasinska (PUT Poznan) Log-time and Log-space Extreme Classification
11:45 - 12:00 Kunal Dahiya (IIT Delhi) Extreme Classification with Label Features
12:00 - 12:15 Xiangru Huang (UT Austin) Dual Decomposed Learning with Factorwise Oracles for Structural SVMs of Large Output Domain
 
12:15 – 13:30 Lunch

Extreme Theory

13:30 - 14:00 Francis Bach (INRIA) Semi-supervised dimension reduction for large numbers of classes
14:00 - 14:15 Scott Yang (NYU) A Theoretical Framework for Structured Prediction using Factor Graph Complexity
14:15 - 14:30 Nagarajan Natarajan (Microsoft) Regret Bounds for Non-decomposable Metrics with Missing Labels
14:30 - 14:45 Cho-Jui Hsieh (UC Davis) Modified GBDTs for Fast Prediction in Extreme Multi-Label Learning
 
14:45 – 15:30 Coffee break

Deep Learning & NLP

15:30 - 16:00 Pascal Vincent (Montréal) Training neural networks in time independent of output layer size
16:00 - 16:15 Edouard Grave (Facebook) Efficient softmax approximation for GPUs
16:15 - 16:30 Stephen Merity (Salesforce) Pointer Sentinel Mixture Models
16:30 - 16:45 Sanjeev Arora (Princeton) A Simple but Tough-to-Beat Baseline for Sentence Embeddings
 
16:45 – 17:00 Break

Deep Learning & Vision

17:00 - 17:30 Christoph Lampert (IST Austria) iCaRL: incremental Classifier and Representation Learning
17:30 - 17:45 Tom Zahavy (Technion) Is a picture worth a thousand words? a Deep Multi Modal Product Classification Architecture for e-commerce
17:45 - 18:15 Armand Joulin (Facebook) Learning to Solve Vision without Annotating Millions of Images