Information fusion deals with reconstructing objects from multiple, possibly incomplete and inconsistent observations. The task of scalable information fusion is critical for interdisciplinary research where a comprehensive picture of the subject requires large amounts of data from disparate data sources. Despite its increasing availability, making sense of such data is not trivial. In this talk I will elaborate on challenges... Read More

The web enables transmission of knowledge at a speed and breadth unprecedented in human history, which has had tremendous positive impact on the lives of billions of people. While benign users try to keep the web safe and usable, malicious users add and spread harmful content, manipulate information and twist things in their favor. Having malicious users and their content... Read More

Today’s main-memory databases can support very high transaction rate for OLTP applications. However, when a large number of concurrent transactions contend on the same data records, the system performance can deteriorate significantly. This is especially the case when scaling transaction processing with optimistic concurrency control (OCC) on multicore machines. In this paper, we propose a new concurrency-control mechanism, called transaction... Read More

Matrix-parametrized models, including multiclass logistic regression and sparse coding, are used in machine learning (ML) applications ranging from computer vision to computational biology. When these models are applied to large scale ML problems starting at millions of samples and tens of thousands of classes, their parameter matrix can grow at an unexpected rate, resulting in high parameter synchronization costs that greatly... Read More

This week, we will have one student from Professor Christos Faloutsos' group and Professor Andy Pavlo's group to give a short talk on the on-going research. Then we will have round table discussion of the on-going work of the other members of DB group with the visiting students attending the CSD Open House. Read More

We propose a dynamic topic model for monitoring temporal evolution of market competition by jointly leveraging tweets and their associated images. For a market of interest (e.g. luxury goods), we aim at automatically detecting the latent topics (e.g. bags, clothes, luxurious) that are competitively shared by multiple brands (e.g. Burberry, Prada, and Chanel), and tracking temporal evolution of the brands'... Read More

This week, we will have round table discussion. We will talk about on-going research, and paper submissions. Read More

In this talk I will first give a brief overview of Petuum which encompasses a set of distributed machine learning principles as well as our open-sourced implementations. By discussing the the high level ideas and performance highlights, I hope to show that Big ML systems can benefit greatly from ML-rooted statistical and algorithmic insights. In the second part I will... Read More

Traditional sketches, such as the Bloom filter, the CountMin sketch, and the Space-Saving sketch, estimate set membership, frequency counts, or moments of scalar random variables. In this paper, we extend these approaches to a new family of sketches that approximate moments of vectorial random variables that satisfy convex polytope constraints. One application is the Semidefinite sketch, a succinct way to... Read More

Finding previously unknown patterns that frequently occur on time series is a core task of mining time series. These patterns are known as time series motifs and are essential to associate events and meaningful occurrences within the time series. In this work we propose a method based on a trie data structure, that allows a fast and accurate time series... Read More