Loading Events

← Back to Events

WEB, Room 236-C

+ Google Map
Wisenbaker Engineering Building
http://ecencesg.wpengine.com

Upcoming Events

December 2017
Free

SPECIAL CESG SEMINAR: Four CESG Graduate Students Present! – Yingyezhe Jin

December 1 @ 4:10 am - 5:10 pm

“Calcium-Modulated Supervised Spike-Timing-Dependent Plasticity for Readout Training and Sparsification of the Liquid State Machine” Abstract: The Liquid State Machine (LSM) is a promising model of recurrent spiking neural networks. It consists of a fixed recurrent network, or the reservoir,  which  projects to a readout layer through plastic readout synapses. The classification performance is highly dependent on the training of readout synapses which tend to be very dense and contribute significantly to the overall network complexity. We present a unifying biologically inspired calcium-modulated supervised spike-timing-dependent plasticity (STDP) approach to training and sparsification of readout synapses, where  supervised temporal learning is modulated  by the post-synaptic firing level characterized by the post-synaptic calcium concentration. The proposed approach prevents synaptic weight saturation, boosts learning performance,  and sparsifies the connectivity between the reservoir and readout layer.  Using the recognition rate of spoken English letters adopted from the TI46 speech corpus as a measure of performance, we demonstrate that the proposed approach outperforms a baseline supervised STDP mechanism by up to 25%, and a competitive non-STDP spike-dependent training algorithm by up to 2.7%. Furthermore, it can prune out up to 30% of readout synapses without causing significant performance degradation. Biography: Yingyezhe Jin is a fifth year Ph.D. candidate in the Department of Electrical and Computer Engineering, Texas A&M University. His current research interests include machine learning and training algorithms for spiking neuron networks. He received a B.S. degree in Electronic and Information Engineering from Zhejiang University, Hangzhou, China, in 2013. 

Find out more »
Free

SPECIAL CESG SEMINAR: Four CESG Grad Students Present! – Abbas Fairouz

December 1 @ 4:10 pm - 5:10 pm

“A Novel Hardware Hash Unit Design for use in Modern Microprocessors” Abstract: Historically, microprocessor instructions were designed in order to obtain high performance on integer and floating point computations. Today’s applications, however, demand high performance for cloud computing, web-based search engines, network applications, and social media tasks. Such software applications involve an extensive use of hashing in their computation. Hashing can reduce the complexity of search and lookup from O(n) to O(n/k), where k bins are used. In modern microprocessors, hashing is done in software. In this work, we propose a novel hardware hash unit design for use in modern microprocessors. We present the design of the Hash Unit (HU) at the micro-architecture level. We simulate the new HU to compare its performance with a software-based hash implementation. We demonstrate a significant speed-up (up to 13x) for the HU. Furthermore, the performance scales elegantly with increasing database size and application diversity, without increasing the hardware cost. Biography:

Find out more »
Free

SPECIAL CESG SEMINAR: Four CESG Grad Students Present! – Vamseedhar Reddyvari

December 1 @ 4:10 pm - 5:10 pm

“Mode-Suppression: A Simple and Provably Stable Chunk-Sharing Algorithm for P2P Networks” Abstract: The ability of a P2P network to scale its throughput up in proportion to the arrival rate of peers has recently been shown to be crucially dependent on the chunk sharing policy employed. Some policies can result in low frequencies of a particular chunk, known as the missing chunk syndrome, which can dramatically reduce throughput and lead to instability of the system. For instance, commonly used policies that nominally “boost” the sharing of infrequent chunks such as the well-known rarest-first algorithm have been shown to be unstable. Recent efforts have largely focused on the careful design of boosting policies to mitigate this issue.  We take a complementary viewpoint, and instead consider a policy that simply prevents the sharing of the most frequent chunk(s). Following terminology from statistics wherein the most frequent value in a data set is called the mode, we refer to this policy as mode suppression. We prove the stability of this algorithm using Lyapunov techniques. We also design a distributed version that suppresses the mode via an estimate obtained by sampling three randomly selected peers. We show numerically that both algorithms perform well at minimizing total download times with distributed mode suppression outperforming all others that we tested against. Biography: Vamseedhar Reddyvari is a PhD student in Electrical & Computer Engineering at Texas A&M University. He got his bachelor’s from IIT Bombay and master’s from IIT Kanpur both in Electrical Engineering. His area of interests are in Networking, Game Theory and Stochastic Learning.

Find out more »
+ iCal Import Listed Events