Extreme classification, where one needs to deal with multi-class and multi-label problems involving an extremely large number of labels, has opened up a new research frontier in machine learning. Many challenging applications, such as photo, video and tweet annotation and web page categorization, can benefit from being formulated as supervised learning tasks with millions of labels. Extreme classification can also lead to a fresh perspective on other learning problems such as ranking and recommendation by reformulating them as multi-class/label tasks where each item to be ranked or recommended is a separate label.
Extreme classification raises a number of interesting research questions including those related to:
* Large scale learning and distributed and parallel training
* Log-time and log-space prediction and prediction on a test-time budget
* Label embedding and tree approaches
* Crowd sourcing, preference elicitation and other data gathering techniques
* Bandits, semi-supervised learning and other approaches for dealing with training set biases and label noise
* Bandits with an extremely large number of arms
* Fine-grained classification
* Zero shot learning and extensible output spaces
* Tackling label polysemy, synonymy and correlations
* Structured output prediction and multi-task learning
* Learning from highly imbalanced data
* Dealing with tail labels and learning from very few data points per label
* PU learning and learning from missing and incorrect labels
* Feature extraction, feature sharing, lazy feature evaluation, etc.
* Performance evaluation
* Statistical analysis and generalization bounds
* Applications to ranking, recommendation, knowledge graph construction and other domains
The workshop aims to bring together researchers interested in these areas to foster discussion and improve upon the state-of-the-art in extreme classification. We also aim to bring researchers from the recommender systems, information retrieval, data mining and computer vision communities to discuss real world application scenarios, evaluation metrics, best practices, etc. Several leading researchers will present invited talks detailing the latest advances in the field. We also seek extended abstracts presenting work in progress which will be reviewed for acceptance as spotlight+poster or a talk. The workshop should be of interest to researchers in core supervised learning as well as application domains such as recommender systems, computer vision, computational advertising, information retrieval and natural language processing. We expect a healthy participation from both industry and academia.
The workshop venue is Room 514a on level 5 of the NIPS convention centre. Participants at this year's workshop will have an option of staying back during the lunch break for John Langford's Vowpal Wabbit 8.1 tutorial from 13:15 - 14:15. This is an additional feature not directly related to the workshop, though John might touch upon extreme multi-class algorithms, and participants are free to choose between lunch and the tutorial (fast eaters can choose both!)
08:00 - 08:10 Introduction
Manik Varma and Moustapha Cisse
Extreme Multi-class Classification
08:10 - 08:40 Samy Bengio (Google)
Sharing is Caring in the Land of The Long Tail
08:40 - 09:10 Yiming Yang (CMU)
Large-scale Structured Learning for Statistical Classification
09:10 - 09:40 Paul Mineiro (Microsoft)
Eigenpartition Trees for Extreme Classification
09:40 - 10:10 Panel discussion
with Samy, Yiming & Paul chaired by John Langford
10:10 - 10:30 Coffee break
Extreme Theory
10:30 – 11:00 Marius Kloft (HU Berlin)
Multi-class SVMs: From Tighter Data-Dependent Generalization Bounds to Novel Algorithms
11:00 – 11:30 Krzysztof Dembczynski (UT Poznan)
Extreme F-measure Maximization
11:30 – 12:00 Ohad Shamir (Weizmann Institute)
Multiclass-Multilabel Classification with More Classes than Examples
12:00 – 12:30 Panel discussion
with Marius, Krzysztof and Mehryar Mohri chaired by Prateek Jain
12:30 – 14:30 Lunch
with an option to attend John Langford's Vowpal Wabbit 8.1 tutorial from 13:15 - 14:15.
Performance Evaluation in Long Tail Applications
14:30 – 15:00 Asela Gunawardana (Microsoft)
Evaluating Machine Learned User Experiences
15:00 – 15:30 Noam Koenigstein (Microsoft)
Implicit Feedback and Performance Evaluation in Recommender Systems
15:30 – 16:00 Panel discussion
with Asela, Noam, Armand Juolin and Patrice Simard chaired by Manik Varma
16:00 – 16:30 Coffee break
Extreme Multi-label Learning
16:30 – 17:00 Charles Elkan (Amazon/UCSD)
Massive Sparse Multilabel Learning
17:00 – 17:30 David Belanger (UMass)
Scaling up Multilabel Classification using Structured Prediction Energy Networks
17:30 – 18:00 Michiel Stock (Ghent)
A two-step method to incorporate task features for large output spaces
18:00 – 18:30 Panel discussion
with Charles, Manik, David and Michiel chaired by Moustapha Cisse
http://research.microsoft.com/en-us/um/people/manik/events/xc15/
