Graduate

OSA Special Seminar

Topic: 
Fluorescence Lifetime Spectroscopy and Imaging for Biomedical Diagnostics
Abstract / Description: 

Fluorescence measurements can provide information about changes in the biochemical, functional and structural characteristics of fluorescent bio-molecular complexes in tissues and cells (e.g. structural proteins, enzyme metabolic co-factors, lipid components, and porphyrins). Typically, these changes are a result of either pathological transformations or therapeutic interventions. We research the development of instrumentation that utilizes label-free fluorescence lifetime contrast to detect and evaluate these molecular changes in vivo in patients and methodologies conducive to near-real time diagnosis of tissue pathologies. This presentation overviews clinically-compatible multispectral fluorescence lifetime imaging techniques developed in our laboratory and their ability to operate as stand-alone diagnostic devices, integrated in a biopsy needle and in conjunction with the da Vinci surgical robot. We present pre-clinical and clinical studies in patients that demonstrate the potential of these techniques for intraoperative assessment of brain tumors and head and neck cancer and intravascular evaluation of atherosclerotic cardiovascular disease. Current results demonstrate that intrinsic fluorescence signals can provide useful contrast for delineation distinct types of tissues including tumors intraoperatively. Challenges and solutions in the clinical implementation of these techniques are discussed.

Date and Time: 
Thursday, April 30, 2015 - 4:15pm to 5:15pm
Venue: 
Physics & Astrophysics Building (HEPL) Rm 102/103

EE380 Computer Systems Colloquium

Topic: 
Knowledge Vault and Knowledge-Based Trust
Abstract / Description: 

In this talk we describe our knowledge extraction and fusion efforts at Google, including the Knowledge Vault project and the Knowledge-based Trust project. We use 15 extractors to periodically extract knowledge from 1B+ Webpages. The results are 3B+ distinct (subject, predicate, object) knowledge triples. Errors can creep in at every stage in this process, both from erroneous data provided by the Web sources and from mistakes made by the extractors. As a result, only about 20% of the extracted triples are correct.

We adapt state-of-the-art data fusion techniques to solve the knowledge fusion problem. By leveraging the collective wisdom from different extractors and from different Web sources, we are able to compute well-calibrated probabilities for the correctness of each triple as well as the correctness of extractions. In addition, we are able to compute trustworthiness for 119M webpages and 5.6M websites. We discuss our observations and provide insights on future research directions.

Date and Time: 
Wednesday, April 29, 2015 - 4:15pm to 5:15pm
Venue: 
Gates B01

SystemX Seminar

Topic: 
Power and Sensor Semiconductors Drive Automotive Applications
Abstract / Description: 

Cars are increasingly driven by electronics to reduce human error, improve traffic flow and to meet environmental regulations. The semiconductor components that enable this functionality range from medium voltage discretes replacing relays to integrated, high-voltage motor drivers with re-programmability at high temperature. In this talk we will review the technology trends underlying the improvements in power discretes, such as IGBTs and GaN HEMT devices, the scaling trends and integration needs of high-voltage BCD CMOS flows, as well as the adjacent assembly challenges of power devices and power integrated modules.

Date and Time: 
Thursday, May 7, 2015 - 4:00pm to 5:00pm
Venue: 
Packard 202

IT-Forum

Topic: 
Social Learning in Decision-Making Groups
Abstract / Description: 

People have always been influenced by the opinions of their acquaintances. Increasingly, through recommendations and ratings provided on all sorts of goods and services, people are also influenced by the opinions of people that are not even acquaintances. This ubiquity of the sharing of opinions has intensified the interest is the concept of herding (or informational cascades) introduced in 1992. While agents in most previous works have only individualistic goals, this talk focuses on social influence among agents in two collaborative settings.

We consider agents that perform Bayesian binary hypothesis testing and, in addition to their private signals, observe the decisions of earlier-acting agents. In the first setting, each decision has its own corresponding Bayes risk. Each agent affects the minimum possible Bayes risk for subsequent agents, so an agent may have a mixed objective including her own Bayes risk and the Bayes risks of subsequent agents; we demonstrate her tension between being informative to other agents and being right in her own decisions, and we show that she is more informative to others when she is open minded. In the second setting, opinions are aggregated by voting, and all agents aim to minimize the Bayes risk of the team's decision. We show that social learning is futile when the agents observe conditionally independent and identically distributed private signals (but not merely conditionally independent private signals) or when the agents require unanimity to make a decision. Our experiments with human subjects suggest that when opinions of people with equal qualities of information are aggregated by voting, the ballots should be secret. They have also raised questions about rationality and trust.

Date and Time: 
Friday, May 1, 2015 - 1:00pm to 2:00pm
Venue: 
Packard 202

SystemX Seminar

Topic: 
Minimally-Invasive Neural Interfaces: From Physical Implants to Virtual Implants
Abstract / Description: 

The development of new neural technologies will revolutionize our understanding of the brain function enabling us to mitigate nervous system disorders and also advance brain-machine interfaces (BMIs).

In this talk, I will first discuss our next generation hybrid parylene-silicon implantable optrodes for neural recording and stimulation. Benefiting from the best of both silicon and polymer material properties, these probes have greater than ten times the density and are two orders of magnitude more compliant than the state of the art, exerting minimal tissue damage and tethering forces while still providing full-volume cortical sampling. I will also introduce a novel parylene-in-parylene (PiP) photonics platform that can be monolithically integrated with our probes for high-resolution optogenetic stimulation.

I will also touch upon a recent radical complementary approach we have devised to guide and steer light in the brain for targeted optogenetic stimulation. In this method, we use non-invasive ultrasound to effectively guide and steer light without physically implanting either a waveguide or a light source. To reach brain structures >2 mm deep, we are developing tetherless upconverting nanocrystal light bulbs that absorb guided near-infrared light and emit visible light to locally stimulate opsins.

Date and Time: 
Thursday, April 30, 2015 - 4:00pm to 5:00pm
Venue: 
Packard 202

ISL Colloquium

Topic: 
The Art of Sequential Optimization via Simulations
Abstract / Description: 

I will talk about a natural framework for simulation-based optimization and control of Markov decision process (MDP) models. The idea is very simple: Replace the Bellman operator by its 'empirical' variant wherein Expectation is replaced by a sample average approximation. This leads to a random Bellman operator in the dynamic programming equations. We introduce several notions of probabilistic fixed points of such random operators, and show their asymptotic equivalence. We establish convergence of empirical Value and Policy Iteration algorithms by a stochastic dominance argument. The mathematical technique introduced is useful for analyzing other iterated random operators (than just the empirical Bellman operator), and may also be useful in analyzing other algorithms for stochastic optimization. The idea can be generalized to asynchronous dynamic programming, and is also useful for computing equilibria of zero-sum stochastic games. Preliminary numerical results show better convergence rate and runtime performance than stochastic approximation/reinforcement learning, or any other commonly used schemes. I will end by talking about Empirical Q-Value Iteration (EQVI), an alternative method for reinforcement learning (RL) with better performance in practice than popular RL algorithms.

This is joint work with Dileep Kalathil, William Haskell and Vivek Borkar.

Date and Time: 
Thursday, April 30, 2015 - 4:15pm to 5:15pm
Venue: 
Packard 101

EE Special Seminar

Topic: 
Power / Performance Optimization for High-End Microprocessors
Abstract / Description: 

A Classic ATE testing focuses on finding defective units and, eliminating them. This talk is about the good units that are left – Bin Split 101. There will be a discussion of manufacturing flow, distributions, and optimizations using testing.

Date and Time: 
Thursday, April 30, 2015 - 4:00pm to 5:30pm
Venue: 
Gates 415

GSEE (Graduate Students in Electrical Engineering) Event

Topic: 
How to Work with Business People
Abstract / Description: 

As a startup founder your success depends largely on the strength and diverse skill set of your team, and whether you are able to work with them effectively. Alan will highlight best practices in operating rhythm between business people and engineers.

Alan is a partner at XSeed Capital and invests in seed-stage enterprise startups. He has 15 years of experience in engineering and product management in enterprise IT companies. Alan holds a BASc in Electrical and Computer Engineering from the University of British Columbia, as well as an MS in Management from the Stanford Graduate School of Business, where he was a Sloan Fellow focused on entrepreneurship, venture capital, and mergers and acquisitions.

This event is hosted by:

  • Graduate Students in Electrical Engineering (GSEE)
  • Graduate Student Council (GSC)
  • Center for Entrepreneurial Studies (CES)

RSVP to Attend

Date and Time: 
Thursday, April 23, 2015 - 6:30pm to 8:00pm
Venue: 
Packard 101

GSEE (Graduate Students in Electrical Engineering) Event

Topic: 
How to Work with Business People
Abstract / Description: 

As a startup founder your success depends largely on the strength and diverse skill set of your team, and whether you are able to work with them effectively. Alan will highlight best practices in operating rhythm between business people and engineers.

Alan is a partner at XSeed Capital and invests in seed-stage enterprise startups. He has 15 years of experience in engineering and product management in enterprise IT companies. Alan holds a BASc in Electrical and Computer Engineering from the University of British Columbia, as well as an MS in Management from the Stanford Graduate School of Business, where he was a Sloan Fellow focused on entrepreneurship, venture capital, and mergers and acquisitions.

This event is hosted by:

  • Graduate Students in Electrical Engineering (GSEE)
  • Graduate Student Council (GSC)
  • Center for Entrepreneurial Studies (CES)

RSVP to Attend

Date and Time: 
Thursday, April 23, 2015 - 6:30pm to 8:00pm
Venue: 
Packard 101

SmartGrid Seminar: Energy Efficiency: The Next Level

Topic: 
Energy Efficiency: The Next Level
Abstract / Description: 

Governor Brown and California's legislative leadership have called for a doubling of energy savings in existing buildings through 2030. In order to achieve these landmark goals and attract sufficient investment, major innovations are needed in energy efficiency policies, markets, and technology. This presentation summarizes a new initiative at Stanford University examining the framework needed for this "Next Level of EE" and a just released Stanford draft report, "Challenges, Opportunities, and New Tools for The Next Level of Energy Efficiency". A particular focus of the presentation will be the role of energy efficiency in a changing electric grid.

Date and Time: 
Thursday, April 23, 2015 - 1:00pm to 2:00pm
Venue: 
Y2E2 300

Pages

Subscribe to RSS - Graduate