Speaker: Adam Mashiach,

M.Sc. student under the supervision of Prof. Ram Zamir.

Wednesday, December 7^th, 2016 at 15:00

Room 011, Kitot Bldg., Faculty of Engineering

Two Asymmetric Descriptions from Many Symmetric Descriptions

Abstract

Multiple descriptions provide a mechanism for graceful degradation in open-loop lossy source coding over a lossy-packet network. Existing solutions for the general asymmetric case are, however, cumbersome. We present a framework for designing an efficient asymmetric two-description scheme, based on dividing many symmetric descriptions into two groups, and encoding each group jointly.

In the first part, we present a multiple description coding scheme which is based on oversampling and dithered delta-sigma quantization that can generate large number of symmetric descriptions.  In the second part, we use this symmetric many description scheme to build an asymmetric sum-rate optimal two-description scheme. We further show that this idea can be extended to generate any number asymmetric descriptions (not only two), which are sum-rate optimal for an interesting specific case of the many description problem.

 (The talk will be given in English)

Speaker:     Dr. Tomer Michaeli
                   EE, Technion, Haifa

Monday, December 5^th, 2016
15:00 - 16:00

Room 011, Kitot Bldg., Faculty of Engineering

Nonparametric Canonical Correlation Analysis

Abstract

Canonical correlation analysis (CCA) is a classical representation learning technique for finding correlated variables in multi-view data. This tool has found widespread use in various fields, including recent application to natural language processing, speech recognition, genomics, and cross-modal retrieval. One of the shortcomings of CCA is its restriction to linear mappings, since many real-world multi-view datasets exhibit highly nonlinear relationships. In recent years, several nonlinear extensions of the original linear CCA have been proposed, including kernel and deep neural network methods. These approaches significantly improve upon linear CCA in many practical applications, but have three limitations. First, they are still restricted to families of projection functions which the user must specify (by choosing a kernel or neural network structure). Second, they are computationally demanding. Third, they often produce redundant representations.
In this work, we derive a closed form solution to the nonlinear CCA problem without any functional restrictions. As we show, the solution corresponds to the SVD of a certain operator associated with the joint density of the views. Thus, by estimating the population density from training data, we obtain a practical nonparametric CCA (NCCA) algorithm, which reduces to solving an eigenvalue system. Superficially, this is similar to kernel CCA, but importantly, NCCA does not require the inversion of any kernel matrix. We also derive a partially linear CCA (PLCCA) variant in which one of the views undergoes a linear projection while the other is nonparametric. Finally, we show why spectral representation learning methods may produce redundant representations, and propose a generic technique to prevent redundancy in those algorithms.
As we demonstrate on several test cases, our algorithms are memory-efficient, often run much faster and perform better than kernel CCA and comparable to deep CCA.

This is joint work with Weiran Wang, Karen Livescu and Yochai Blau.

קול קורא לבוגרי הפקולטה - בואו לשמוע ולהשפיע

מעבדת CIM  - computer integrated manufacturing

מארחת את בוגרי הפקולטה

יום רביעי 04.01.17 בין השעות 18:30-20:30

המפגש מיועד לבוגרי הפקולטה להנדסה, אוניברסיטת תל אביב בלבד

נדרשת הרשמה מראש

לינק לפרטים והרשמה

במפגש, הצגת המעבדה והנעשה בה עם צוות המעבדה ודוקטורנטים.

כולל דיון ושיח פתוח על מנת להיעזר בניסיון שלכם. לשמוע מהם הנושאים שאתם הייתם ממליצים ללמד במעבדה