Graduate

SmartGrid: Michael Caramanis (Boston University)

Topic: 
Smart grids and energy systems
Abstract / Description: 

LECTURE SERIES - FALL QUARTER OF 2014

Our speakers will discuss exciting new ideas and technologies that are changing the electricity industry. The theme of the seminar series is on smart grids and energy systems, scheduled to be held approximately every Thursday, with speakers from academic institutions and industry. The hour-long seminars, including ample time for discussion, are held at 1 p.m. approximately every other Thursday.

Date and Time: 
Thursday, November 6, 2014 - 1:00pm to 2:00pm
Venue: 
Y2E2 101

SmartGrid: Sila Killicote (Lawrence Berkeley National Laboratory)

Topic: 
Smart grids and energy systems
Abstract / Description: 

LECTURE SERIES - FALL QUARTER OF 2014

Our speakers will discuss exciting new ideas and technologies that are changing the electricity industry. The theme of the seminar series is on smart grids and energy systems, scheduled to be held approximately every Thursday, with speakers from academic institutions and industry. The hour-long seminars, including ample time for discussion, are held at 1 p.m. approximately every other Thursday.

Date and Time: 
Thursday, October 30, 2014 - 1:00pm to 2:00pm
Venue: 
Y2E2 101

SmartGrid: Tim Heidel (Advanced Research Projects Agency - Energy)

Topic: 
Smart grids and energy systems
Abstract / Description: 

LECTURE SERIES - FALL QUARTER OF 2014

Our speakers will discuss exciting new ideas and technologies that are changing the electricity industry. The theme of the seminar series is on smart grids and energy systems, scheduled to be held approximately every Thursday, with speakers from academic institutions and industry. The hour-long seminars, including ample time for discussion, are held at 1 p.m. approximately every other Thursday.

Date and Time: 
Thursday, October 16, 2014 - 1:00pm to 2:00pm
Venue: 
Y2E2 101

SmartGrid: Javad Lavaei (Columbia University)

Topic: 
Smart grids and energy systems
Abstract / Description: 

LECTURE SERIES - FALL QUARTER OF 2014

Our speakers will discuss exciting new ideas and technologies that are changing the electricity industry. The theme of the seminar series is on smart grids and energy systems, scheduled to be held approximately every Thursday, with speakers from academic institutions and industry. The hour-long seminars, including ample time for discussion, are held at 1 p.m. approximately every other Thursday.

Date and Time: 
Thursday, October 9, 2014 - 1:00pm to 2:00pm
Venue: 
Y2E2 101

SmartGrid: Yilu Liu (University of Tennessee)

Topic: 
Smart grids and energy systems
Abstract / Description: 

LECTURE SERIES - FALL QUARTER OF 2014

Our speakers will discuss exciting new ideas and technologies that are changing the electricity industry. The theme of the seminar series is on smart grids and energy systems, scheduled to be held approximately every Thursday, with speakers from academic institutions and industry. The hour-long seminars, including ample time for discussion, are held at 1 p.m. approximately every other Thursday.

Date and Time: 
Thursday, September 25, 2014 - 1:00pm to 2:00pm
Venue: 
Y2E2 101

IT-Forum: What is quantum information theory?

Topic: 
What is quantum information theory?
Abstract / Description: 

Quantum information theory is the generalization of Shannon’s information theory to the realm of quantum mechanics. The basic questions are the same as for standard information theory: What is uncertainty? What is the best method for sending information through a noisy channel? The theory, however, is full of surprises. Random variables don’t commute. Correlation of the quantum variety increases channel capacities. Conditional entropy can be negative. Secrecy can be certified. I’ll give a brief introduction to quantum information theory, highlighting some of the key similarities and striking differences with Shannon theory.

 


The Information Theory Forum (IT-Forum) at Stanford ISL is an interdisciplinary academic forum which focuses on mathematical aspects of information processing. With a primary emphasis on information theory, we also welcome researchers from signal processing, learning and statistical inference, control and optimization to deliver talks at our forum. We also warmly welcome industrial affiliates in the above fields. The forum is typically held in Packard 202 every Friday at 1:00 pm during the academic year.

The Information Theory Forum is organized by graduate students Jiantao Jiao and Kartik Venkat. To suggest speakers, please contact any of the students.

Date and Time: 
Friday, October 31, 2014 - 1:00pm to 2:00pm
Venue: 
Packard 202

IT-Forum: A tale of two uncertainties

Topic: 
A tale of two uncertainties
Abstract / Description: 

We discuss two separate results, related only through the list of authors.

First, we consider bounding the mutual information for distributions p(x,y) with "sparse" supports. To that end, we introduce the notion of "intrinsic uncertainty", which measures the uncertainty remaining as to which "action" has been taken by the channel p(y|x) after observing both its input and its output. Using the Donsker-Varadhan variational principle, we derive a lower bound for the intrinsic uncertainty that depends only on the marginals and support of p(x,y). This bound easily translates into a bound on the mutual information. We apply this method to the binary deletion channel, and show that for the special case of an i.i.d. input, our bound outperforms the best known lower bound for the mutual information over a wide range of deletion probabilities.

Next, we consider the AWGN channel with noisy feedback. Schalkwijk and Kailath have shown that for noiseless feedback, capacity can be attained via a simple scalar linear scheme. This elegant scheme fails however in the presence of arbitrarily low feedback noise, essentially since the uncertainty at the transmitter regarding the receiver's state grows rapidly, decoupling the terminals. In fact, a general negative result of Kim, Lapidoth and Weissman indicates that linear encoding schemes cannot attain any positive rate with asymptotically vanishing error. Is there nevertheless a "simple" scheme robust to feedback noise in some sense? We move away from the asymptotic approach, and present a simple scalar interactive scheme that can operate close to capacity with a finite (but very small) error probability, in a small number of rounds. Our scheme is a variation on the Schakwijk-Kailath theme with active feedback, where transmitter uncertainty growth is curbed using modulo arithmetic. For example, for an error probability of 1e-6, if the feedback SNR exceeds the forward SNR by 20dB, our scheme operates 0.8dB from capacity with only 19 rounds of interaction. In comparison, error correcting codes attaining the same performance require two orders of magnitude increase in delay and complexity, and a minimal delay/complexity uncoded system is bounded 9dB away from capacity.

Joint work with Or Ordentlich and Assaf Ben-Yishai.

 

The Information Theory Forum (IT-Forum) at Stanford ISL is an interdisciplinary academic forum which focuses on mathematical aspects of information processing. With a primary emphasis on information theory, we also welcome researchers from signal processing, learning and statistical inference, control and optimization to deliver talks at our forum. We also warmly welcome industrial affiliates in the above fields. The forum is typically held in Packard 202 every Friday at 1:00 pm during the academic year.

The Information Theory Forum is organized by graduate students Jiantao Jiao and Kartik Venkat. To suggest speakers, please contact any of the students.

Date and Time: 
Friday, October 3, 2014 - 1:00pm to 2:00pm
Venue: 
Packard 202

IT-Forum: Jiaming Xu (UIUC)

Topic: 
Fundamental Limits for Community Detection
Abstract / Description: 

We consider a generative model for a network, namely the planted cluster model, which is a simple extension of the classical Erdos-Renyi random graph. The task is to exactly recover the planted clusters from observation of the network.

We first derive an information limit for exact cluster recovery and show it is achieved by the maximum likelihood (ML) estimation up to constant factors. We then show a convex relaxation of ML estimation can successfully recover the clusters up to a spectral barrier but fails to achieve the information limit. We conjecture no polynomial-time algorithm can succeed significantly beyond the spectral barrier and achieve the information limit. To provide evidence, we show recovering a single cluster significantly beyond the spectral barrier is at least as hard as detecting a clique of size o(sqrt(n)) planted in an Erdos-Renyi random graph with n nodes and a constant edge probability.

This work is at the intersection of information theory, machine learning, and high-dimensional statistics. Based on joint work with Yudong Chen at UC Berkeley, and Bruce Hajek, Yihong Wu from UIUC.

Date and Time: 
Friday, September 12, 2014 - 1:00pm
Venue: 
Packard 202

SCIEN Talk: Quanta Image Sensor (QIS) Concept and Progress

Topic: 
Quanta Image Sensor (QIS) Concept and Progress
Abstract / Description: 

The Quanta Image Sensor (QIS) was conceived when contemplating shrinking pixel sizes and storage capacities, and the steady increase in digital processing power. In the single-bit QIS, the output of each field is a binary bit plane, where each bit represents the presence or absence of at least one photoelectron in a photodetector. A series of bit planes is generated through high speed readout, and a kernel or “cubicle” of bits (X,Y, t) is used to create a single output image pixel. The size of the cubicle can be adjust post-acquisition to optimize image quality. The specialized sub-diffraction-limit photodetectors in the QIS are referred to as “jots” and a QIS may have a gigajot or more, read out at 1000 fps, for a data rate exceeding 1Tb/s. Basically, we are trying to count photons as they arrive at the sensor. Recent progress towards realizing the QIS for commercial and scientific purposes will be discussed. This includes investigation of a pump-gate jot device implemented in a 65nm process, power efficient readout electronics, currently less than 20pJ/b in 0.18 um CMOS, creating images from jot data with high dynamic range, and understanding the imaging characteristics of single-bit and multi-bit QIS devices, such as the inherent and interesting film-like D-log(H) characteristic. If successful, the QIS will represent a major paradigm shift in image capture.

Date and Time: 
Wednesday, October 1, 2014 - 4:15pm to 5:15pm
Venue: 
Packard 101

Pages

Subscribe to RSS - Graduate