TECHNICAL SESSIONS

Session papers are available to workshop registrants immediately and to everyone beginning December 11, 2008.

Opening Remarks

Program Co-Chairs: Armando Fox, University of California, Berkeley; Sumit Basu, Microsoft Research

Invited Talk

Patterns (and Anti-Patterns) for Developing Machine Learning Systems
Gordon Rios, Chief Technology Officer, Zvents, Inc.

View the presentation slides

Design patterns are a staple in the software engineering field and have helped developers capture and reuse successful approaches to common problems. As the field of machine learning continues to mature, it may be helpful to discuss some of the implementation patterns that can lead to successful deployments. We'll also touch on some anti-patterns, a much more controversial topic, where a seemingly obvious approach can lead to less successful outcomes. Rather than a manifesto or final word, my intent is to provide a starting point for the conversation.

Bio: Gordon Rios is a proven leader in search and search-related technologies. Over the past 10 years he has focused his efforts on devising machine learning systems for Web search, document classification, and text mining. At Inktomi he developed the scoring engine for a user click-based ranking application that was deployed at HotBot and Snap; was a major contributor to the Directory Engine product, which classified millions of Web documents into over 10,000 categories; and was a founding member of Inktomi's Web Search Relevance Group. Moving to Proofpoint, an enterprise messaging company, he led the development of the award-winning MLX™ technology used in their industry-leading anti-spam product. Most recently, Gordon worked at Yahoo! on the International Relevance team, where he worked on all aspects of search engine relevance and developed production code running in all major international markets. Gordon holds an MBA in Finance and a Master's Degree in Engineering and Computer Science from the University of California, Berkeley.

Oral Session 1

Finding Similar Failures Using Callstack Similarity
Kevin Bartz, Harvard University; Jack W. Stokes and John C. Platt, Microsoft Research, Redmond; Ryan Kivett, David Grant, Silviu Calinoiu and Gretchen Loihle, Microsoft Corporation, Redmond

Paper in HTML | PDF

Empirical Comparison of Techniques for Automated Failure Diagnosis
Songyun Duan and Shivnath Babu, Duke University

Paper in HTML | PDF

HiLighter: Automatically Building Robust Signatures of Performance Behavior for Small- and Large-Scale Systems
Peter Bodík, University of California, Berkeley; Moises Goldszmidt, Microsoft Research, Silicon Valley; Armando Fox, University of California, Berkeley

Paper in HTML | PDF

Oral Session 2

Mining Console Logs for Large-Scale System Problem Detection
Wei Xu, University of California, Berkeley; Ling Huang, Intel Research Berkeley; Armando Fox, David Patterson, and Michael Jordan, University of California, Berkeley

Paper in HTML | PDF

An Internet Protocol Address Clustering Algorithm
Robert Beverly and Karen Sollins, MIT CSAIL

Paper in HTML | PDF

Poster Session (with refreshments)

Reinforcement Learning for Optimizing Flash Memory Cache Management
Byung Kon Kang, Kee-Eung Kim, Wook Jung, and Jin-Soo Kim, KAIST

Abstract

Predictive System Health Monitoring for Large-Scale Hosting Infrastructures
Xiaohui Gu, North Carolina State University; Haixun Wang, IBM T.J. Watson Research

Abstract

Ganesha: Black-Box Fault Diagnosis for MapReduce Systems
Xinghao Pan, Jiaqi Tan, Soila Kavulya, Rajeev Gandhi, and Priya Narasimhan, Carnegie Mellon University

Abstract

Using Machine Learning to Auto-tune a Stencil Code on a Multicore Architecture
Archana Ganapathi, Kaushik Datta, Armando Fox, and David Patterson, University of California, Berkeley

Abstract

Generalization of Machine Learning Based Encrypted Traffic Identification: A Comparison of Two Classifiers
Riyad Alshammari and Nur Zincir-Heywood, Dalhousie University

Abstract

Kudzu: A Self-Balancing P2P File Transfer System
Sean Barker, Marius Catalin Iordan, and Jeannie Albrecht, Williams College; Barath Raghavan, University of California, San Diego

Abstract

Oral Session 3

Probabilistic Inference in Queueing Networks
Charles Sutton and Michael I. Jordan, University of California, Berkeley

Paper in HTML | PDF

From Optimization to Regret Minimization and Back Again
Ioannis Avramopoulos, Jennifer Rexford, and Robert Schapire, Deutsche Telekom Laboratories and Princeton University

Paper in HTML | PDF

Wrap-up and Discussion