EE Student Information

Graduate

ISL Colloquium presents AI for clinical trials and clinical trials for AI

Topic: 
AI for clinical trials and clinical trials for AI
Abstract / Description: 

Note: this talk will be held in person in Packard 101, and will start at 4pm. The talk will be streamed on Zoom for those who cannot attend (registration link below).  Please join us for coffee starting at 3:30pm at the Grove outside Packard.

 

 

Date and Time: 
Thursday, October 21, 2021 - 4:00pm
Venue: 
Packard 101 + Zoom

ISL Colloquium presents Two Talks

Topic: 
Experimentation and Decision-Making in Two-Sided Marketplaces: The Impact of Interference / Simple Agent, Complex Environment: Efficient Reinforcement Learning with Agent States
Abstract / Description: 

Talk 1: Marketplace platforms use experiments (also known as "A/B tests") as a method for making data-driven decisions. When platforms consider introducing a new feature, they often first run an experiment to test the feature on a subset of users and then use this data to decide whether to launch the feature platform-wide. However, it is well documented that estimates of the treatment effect arising from these experiments may be biased, due to the presence of interference. In this talk, we survey a collection of recent results and insights we have developed on experimentation and decision-making in two-sided marketplaces. In particular, we study the bias that interference creates in both the treatment effect estimates as well as standard error estimates, and show how both types of biases affect the platform's ability to make decisions. We show that for a large class of interventions ("positive interventions"), these biases cause the platform to launch too often. Through simulations calibrated to real-world data, we show that in many settings the treatment effect bias impacts decision-making more than the standard error bias. Based on joint work with Ramesh Johari, Inessa Liskovich, Gabriel Weintraub, and Geng Zhao.


Talk 2: In this work, we design a simple reinforcement learning (RL) agent that implements an optimistic version of Q-learning and establish through regret analysis that this agent can operate with some level of competence in an arbitrarily complex environment. While we leverage concepts from the literature on provably efficient RL, we consider a general agent-environment interface and provide a novel agent design and analysis. This level of generality positions our results to inform the design of future agents for operation in complex real environments. We establish that, as time progresses, our agent performs competitively relative to policies that require longer times to evaluate. The time it takes to approach asymptotic performance is polynomial in the complexity of the agent's state representation and the time required to evaluate the best policy that the agent can represent. Notably, there is no dependence on the complexity of the environment. The ultimate per-period performance loss of the agent is bounded by a constant multiple of a measure of distortion introduced by the agent's state representation. This work is the first to establish that an algorithm approaches this asymptotic condition within a tractable time frame.

Date and Time: 
Thursday, October 14, 2021 - 4:00pm
Venue: 
Packard 101

Probability Seminar: Nonlinear large deviations: Mean-field and beyond

Topic: 
Nonlinear large deviations: Mean-field and beyond
Abstract / Description: 

Large deviations of nonlinear functions of adjacency matrices of sparse random graphs have gained considerable interest over the last decade. This includes popular examples like subgraph count, or the extreme eigenvalues. For the first half of the talk, we will discuss how the upper tail large deviation problem of subgraph count in a random regular graph can be reduced to a variational problem and how to solve such optimization. Next, we consider Erdos–Renyi graph $\mathcal{G}(n,p)$ in the regime of $p$ where largest eigenvalue is governed by localized statistics, such as high degree vertices. In particular, for $r \geq 1$ fixed, we will discuss the upper and lower tail probabilities of top $r$ eigenvalues jointly.

This talk is based on joint works with Bhaswar B. Bhattacharya, Amir Dembo, and Shirshendu Ganguly.

Date and Time: 
Monday, October 11, 2021 - 4:00pm
Venue: 
Sequoia 200

Clean Energy Education & Empowerment (C3E)

Topic: 
Utility Leadership in Accelerating the Carbon-Free Energy Transition
Abstract / Description: 

Electric utilities across the U.S. are one of the most important players in the nation's efforts to transition to carbon-free electricity and net-zero greenhouse gas emissions. As of September 2021, seventy-two percent of customer accounts in the U.S. are served by an electric utility with a 100 percent carbon-reduction target. Utilities and their customers are deploying clean energy and energy efficiency at a record pace in parallel with retiring fossil fuel plants. This webinar will feature three female CEOs of utilities that are leading the way per the Smart Electric Power Alliance's 2021 "Utility Transformation Challenge". The webinar will focus on the importance of cultural change and deep engagement with stakeholders and partners to reach ambitious carbon-reduction goals.

You can submit your questions in advance of the Q/A portion of the webinar in Slido.

Date and Time: 
Monday, October 25, 2021 - 10:00am
Venue: 
webinar

AP/P Colloquium: "On Ising’s Model of Ferromagnetism"

Topic: 
On Ising’s Model of Ferromagnetism
Abstract / Description: 

The 1D Ising model is a classical model of great historical significance for both classical and quantum statistical mechanics. Developments in the understanding of the Ising model have fundamentally impacted our knowledge of thermodynamics, critical phenomena, magnetism, conformal quantum field theories, particle physics, and fractionalization in many-body systems. Despite the theoretical impact of the Ising model there have been very few good 1D realizations of it in actual real material systems. However, it has been pointed out recently, that the material CoNb2O6, has a number of features that may make it the most ideal realization we have of the Ising model in one dimension. In this talk I will discuss the surprisingly complex physics resulting in this simple model and review the history of “Ising’s model” from both a scientific and human perspective. In the modern context I will review recent experiments by my group and others on CoNb2O6. In particular I will show how low frequency light in the THz range gives unique insight into the tremendous zoo of phenomena arising in this simple model system.

Date and Time: 
Tuesday, October 12, 2021 - 4:30pm

SystemX presents "High Volume Manufacturing of Silicon Spin Qubits"

Topic: 
High Volume Manufacturing of Silicon Spin Qubits
Abstract / Description: 

High-volume manufacturing of semiconductors has enabled the integrations of billions of transistors on a single chip. Intel leverages this expertise to address engineering challenges in the scale-up of quantum computers. In this talk, we discuss how silicon spin qubits are fabricated in a state-of-the-art CMOS 300 mm fabrication line. The qubits are comparable in size to a transistor, and are highly coherent. Further on, advancements in the measurement infrastructure, such as a 300 mm wafer probing tool operating at 1.6 Kelvin, are shown. Lastly, we will discuss how control electronics operating at cryogenic temperatures can address the interconnect challenge all solid state quantum technologies face.

 

Date and Time: 
Thursday, October 14, 2021 - 5:30pm
Venue: 
Huang 018 + Zoom

SCIEN presents "Neural Methods for Reconstruction and Rendering of Real World Scenes"

Topic: 
Neural Methods for Reconstruction and Rendering of Real World Scenes
Abstract / Description: 

In this presentation, I will talk about some of the recent work we did on new methods for reconstructing computer graphics models of real world scenes from sparse or even monocular video data. These methods are based of bringing together neural network-based and explicit model-based approaches. I will also talk about new neural rendering approaches that combine explicit model-based and neural network based concepts for image formation in new ways. They enable new means to synthesize highly realistic imagery and videos of real work scenes under user control.

Date and Time: 
Wednesday, November 17, 2021 - 10:00am

SCIEN presents "Differentiable Simulation of Light"

Topic: 
Differentiable Simulation of Light
Abstract / Description: 

Inverse problems involving light abound throughout many scientific disciplines. Typically, a set of images captured by an instrument must be mathematically processed to reveal some property of our physical reality. This talk will provide an introduction and overview of the emerging field of differentiable physically based rendering, which has the potential of substantially improving the accuracy of such calculations.
Methods in this area propagate derivative information through physical light simulations to solve optimization problems. While still very much a work in progress, advances in the last years have led to increasingly efficient and numerically robust methods that can begin to tackle interesting real-world problems. I will give an overview of recent progress and open problems.

Date and Time: 
Wednesday, November 10, 2021 - 10:00am

SCIEN presents "Freeform Optics for Imaging and Metaform Optics in Near-Eye Displays"

Topic: 
Freeform Optics for Imaging and Metaform Optics in Near-Eye Displays
Abstract / Description: 

Freeform optics has set a new path for optical system design across a wide range of applications spanning from microscopy to space optics, including the billion $ consumer market of near-eye displays for augmented reality that set us on this technology path in the first place. Today, freeform optics have been demonstrated to yield compact, achromatic, and high-performance imaging systems that are poised to enable the science of tomorrow. This talk will introduce freeform optics and highlight emerging design methods. We will then present success stories in digital-viewfinder, imager, and spectrometer designs, which we anticipate will ignite discussion and stimulate cooperation in enabling knowledge in freeform optics. Building on this foundation, we will introduce the concept of a metaform to address a need in near-eye displays.

Date and Time: 
Wednesday, November 3, 2021 - 4:30pm

SCIEN presents "Design Tools for Material Appearance"

Topic: 
Design Tools for Material Appearance
Abstract / Description: 

The design of material appearance for both virtual and physical design remains a challenging problem. There aren’t straightforward intuitive techniques as there are in  geometric design where shapes can be sketched or assembled from geometric primitives. In this talk I will present a series of contributions to developing intuitive appearance design tools. This includes studies of material appearance perception which form the basis of the development of perceptual axes for reflectance distribution design. I will also present novel interfaces for design including hybrid slider/image navigation and augmented reality interfaces. I will discuss the unique problems involved in designing appearance for objects to be physically manufactured rather than simply displayed in virtual environments.  Finally, I will show how exemplars of spatially varying materials can be inverted to produce procedural models.

Date and Time: 
Wednesday, October 27, 2021 - 4:30pm

Pages

Subscribe to RSS - Graduate