| ||||||||||

Gone |
||||||||||

Wednesday, 06.04.08, 11:30am - 1:00pm, 4004/06 CalIT2TBAYoung-Han Kim, ECE, UCSD Shannon's mutual information arises as the canonical answer to a variety of questions such as communication (channel coding theorem), data compression (rate distortion theorem and Gallager's minimax redundancy theorem), and investment (Kelly's growth-optimal portfolio). In this talk, we discuss how causality affects these problems and introduce Massey's directed information as a natural answer...more > | ||||||||||

Wednesday, 05.28.08, 12 - 1.30pm, 4004/4006 (4th floor Atkinson)Entry wise bounds for eigenvectors of random graphsPradipta Mitra, Yale I will talk about a certain spectral property of random graphs. The goal is to investigate what the first eigenvector of the adjacency matrix of a random graph looks like. Let G be a G_{n, p} random graph on n vertices. Also, let A be the adjacency matrix of G, and v_1 be the first eigenvector of A. I will show a nearly optimal bound on |v_1(i) - frac{1}{sqrt{n}}| that holds for all i in [n], with high probability...more > | ||||||||||

Wednesday, 05.07.08, 12pm - 1:00pm, 4004/06 CalIT2Uniform Direct Product Theorems: Simplified, Optimized, and DerandomizedRussell Impagliazzo, CSE, UCSD and IAS The Direct Product Theorem for circuits says, ``If no small circuit can compute Boolean function $f(x)$ with probability at least $1-delta$ for a random $x$, then no circuit of a slightly smaller size can compute $f(x_1)...f(x_k)$ with even a small probability of success.' While highly intuitive, the proof of the direct product theorem is non-trivial...more > | ||||||||||

Wednesday, 04.16.08, 12pm - 1.30pm, 4004/4006 (4th floor Atkinson)Interference Alignment and the Capacity of Wireless NetworksSyed Ali Jafar, UC Irvine We use the idea of interference alignment to characterize the capacity of wireless interference networks with accuracy approaching 100% at high SNR. The talk covers three main results. (1) We show that wireless interference networks are not fundamentally interference limited. At high SNR, inspite of the interference, every user in an interference network can achieve a rate close to 1/2 of his interference-free capacity...more > | ||||||||||

Wednesday, 03.05.08, 12pm - 1.30pm, 4th Floor Conference Room CALIT2Functional SparsityJohn Lafferty, School of Computer Science, Carnegie Mellon University Substantial progress has recently been made on understanding the behavior of sparse linear models in the high dimensional setting, where the number the variables can greatly exceed the number of samples. This problem has attracted the interest of multiple communities, including applied mathematics, signal processing, statistics, and machine learning...more > | ||||||||||

Wednesday, 02.27.08, 12pm - 1.30pm, 4th Floor Conference Room CALIT2Learning Classifiers from Only Positive and Unlabeled DataKeith Noto, UCSD The input to an algorithm that learns a binary classifier consists of two sets of examples. Normally, one set contains positive examples of a concept, and the other set contains negative examples. However, it is often the case that the available input consists of an incomplete set of positive examples, and a set of unlabeled examples, some of which are positive and some of which are negative...more > | ||||||||||

Wednesday, 02.20.08, 12:00pm - 1:00pm, 4004/06 (4th floor conference room), CALIT2Designing a content-based music search engineGert Lanckriet, ECE,UCSD If you go to Amazon.com or Apple Itunes, your ability to search for new music will largely be limited by the `query-by-metadata' paradigm: search by song, artist or album name. However, when we talk or write about music, we use a rich vocabulary of semantic concepts to convey our listening experience. If we can model a relationship between these concepts and the audio content, then we can produce a more flexible music search engine based on a 'query-by-semantic-description' paradigm...more > | ||||||||||

Wednesday, 02.13.08, 12pm - 1.30pm, 4th Floor Conference Room CALIT2Communicating Delay-Sensitive and Bursty Information over an Outage ChannelTara Javidi, UCSD In this talk, we consider the classic (cross-layer) queue-channel optimization problem for bursty and delay-sensitive information sources. In particular, we are interested in communications over outage-limited channels (with no feedback) when the number of bits that arrive at the transmitter during any time slot is random but the delivery of bits at the receiver must adhere to a strict delay constraint...more > | ||||||||||

Wednesday, 02.06.08, 12pm - 1.30pm, 4004/4006 (4th floor Atkinson)Efficient reductions among lattice problemsDaniele Micciancio, CSE, UCSD We give various deterministic polynomial time reductions among approximation problems on point lattices. Our reductions are both efficient and robust, in the sense that they preserve the rank of the lattice and approximation factor achieved. Our main result shows that for any g >= 1, approximating all the successive minima of a lattice (and, in particular, approximately solving the Shortest Independent Vectors Problem, SIVPg) within a factor g reduces under deterministic polynomial time rank-preserving reductions to approximating the Closest Vector Problem (CVP) within the same factor g...more > | ||||||||||

Wednesday, 01.16.08, 12 - 1.30, 4th Floor Conference Room CALIT2Wireless ad-hoc networks: from probability to physics via information theoryMassimo Franceschetti, ECE, UCSD In this interdisciplinary talk we consider the problem of determining the information capacity of a network of wireless transmitters and receivers and try to draw some non-trivial connections between spatial stochastic processes, physics, and information theory. We present the following main result of statistical physics flavor: By distributing uniformly at random an order of n nodes wishing to establish pair-wise independent communications inside a domain of size of the order of n, the per-node information rate must follow an inverse square-root of n law, as n tends to infinity...more > | ||||||||||

Tuesday, 06.12.07, 1 - 2, CSE 4140A Topic in Coding TheoryRoxana Smarandache, San Diego State University | ||||||||||

Tuesday, 06.05.07, 1 - 2, CSE 4140Antenna Synthesis and Channel CapacityMarco Migliore In this lecture the problem of antenna synthesis is reviewed from an information-theoretic perspective. Antennas are characterized by their ability to convey information in a communication system and different classes of antennas, e.g. “classic” antennas, adaptive antennas and MIMO antennas, are analysed in a unified framework using the concept of the number of degrees of freedom of the field...more > | ||||||||||

Friday, 06.01.07, 1 - 2, CSE 2109A physical approach to multiple antenna communicationMassimo Franceschetti, UCSD In multiple antenna (MIMO) systems communication is performed through the act of propagation of electromagnetic (EM) waves. EM research typically focuses on the physical aspects of propagation, while information theory (IT) focuses mainly on the communication aspects, often considering random channel models. In this talk we attempt to address the gap between these two approaches...more > | ||||||||||

Tuesday, 05.29.07, 1 - 3.30, CSE 4140A Stochastic Control Approach to Variable Length Menus in P300 Neural Communication ProsthesesTodd Coleman, University of Illinois at Urbana-Champaign The P300 neural communication prosthesis allows an individual to type words from an on-screen menu by recording visually evoked potentials related to the user’s intention. Typing rates are affected in part by two sources of uncertainty: (a) noise in the P300 signal, and (b) statistical language structure. Early system designs focused on classification to address P300 noise...more > | ||||||||||

Wednesday, 05.23.07, 1 - 2, CSE 2109Distance metric learning for large margin nearest neighbor classificationLawrence Saul, UCSD The accuracy of k-nearest neighbor (kNN) classification depends in general on the metric used to compute distances between different inputs. We show how to learn a Mahanalobis distance metric for kNN classification from labeled examples. The metric is trained with the goal that the k-nearest neighbors always belong to the same class while examples from different classes are separated by a large margin...more > | ||||||||||

Tuesday, 04.17.07, 12 - 1, CSE 4140Using Random Projections to Learn the Distribution of Data Concentrated on a Low Dimensional ManifoldYoav Freund, UCSD In this lunchtime informal seminar, Yoav will discuss some recent work and present some open problems. | ||||||||||

Tuesday, 04.10.07, 12 - 1, CSE 4140Repeat classification and fragment assembly. Pavel Pevzner, UCSD In this inagural lunchtime seminar, Pavel will discuss his research and present some open problems. | ||||||||||

Monday, 10.30.06, 12 - 1, 4004/4006 (4th floor Atkinson)The physical layer of WiMedia OFDM UWBRabih Chrabieh The WiMedia Ultra-Wideband system uses OFDM, rather than ultra-narrow impulses, to fill up a vast amount of spectrum, more than 500 MHz, with high data rate transmissions at short distances, and very low power. In a nutshell, it is like a high data rate Bluetooth. We will present the physical layer of this standard with some references to the MAC layer...more > | ||||||||||