Graduate

Concentration of measure is a collection of tools and results from analysis and probability theory that have been used in many areas of pure and applied mathematics. Arguably, the first data science application of measure concentration (under the name ''blowing-up lemma'') is the proof of strong converses in multiuser information theory by Ahlswede, G\'acs and K\"orner in 1976. Since then, measure concentration has found applications in many other information theoretic problems, most notably the converse (impossibility) results in information theory. Motivated by this, information theorists (e.g. Marton) have also contributed to the mathematical foundations of measure concentration using their information-theoretic techniques.

Now, after all the past 40 years of such progress, we found that, amusingly, measure concentration is not the right hammer for many of these information theoretic applications. We introduce a new machinery based on functional inequalities and reverse hypercontractivity which yields strict improvements in terms of sharpness of the bounds, generality of the source/channel distributions, and simplicity of the proofs. Examples covered in the talk include: 1. optimal second-order converse to common randomness generation with rate-limited communication; 2. sharpening the relay channel converse bounds by and Wu and Ozgur with much simpler proofs.

The work benefited from collaborations with Thomas Courtade, Paul Cuff, Ayfer Ozgur, Ramon van Handel, and Sergio Verd\'u.

A novel analysis of the CMB data from the Planck satellite appears to reveal 20 anomalous pointlike primordial energy sources in the very early universe. Though hard to accommodate within standard inflationary cosmology, such anomalous points are a prediction of conformal cyclic cosmology (CCC), as the conformally compressed Hawking radiation, whereby supermassive black holes in an aeon prior to ours finally disgorge all their mass-energy. The basic features of the CCC model will be described in the lecture, as will a remarkable connection with the 2014 BICEP2 observations.

Sponsored by Bioengineering; of interest to the Physics community.

Emerging trends in applications of communication and data management are driving exciting challenges in the telecom/datacom industry. From 5G driving optical-wireless convergence all the way to hyperscale datacenters bringing the optical signal closer to the switch ASIC, breakthroughs in the technology enabling optical connectivity are becoming critical for the future. In this discussion, we will provide a quick overview of recent evolution and some emerging challenges in advanced optical connectivity, as well as insight on how Corning is embracing the opportunity to solve those challenges with glass.

It is our distinct pleasure to welcome noted hacker and engineer, Andrew "bunnie" Huang

Please join us for his talk, Chibitronics - Democratizing Tech and Bridging the Gender Gap

In this talk, bunnie will discuss some of the technical aspects behind the introduction of paper based electronics, and some process details used to manufacture at scale.

In addition, he will present initial results indicating that, by changing materials and approach, a broader swath of society can become engaged with technology.

Outliers in weights and activations often prevent us from benefiting from very low precision, e.g., 2 or 4 bits. In this talk, we will present our recent works on hardware accelerator and low memory-cost training where we handle a very small amount of outliers, e.g., large values occupying only 1% of total data, in high precision and the majority of data in very low precision.

The accelerator, based on 4-bit computation, offers average 30% better performance and energy efficiency compared with the zero-skipping 8-bit accelerator. The low memory-cost training solution applies 3-bit precision to 98% of activations to be stored during training, which finally leads to 9X reduction in the memory cost of ResNet-152 while keeping the training accuracy.

End of Summer Quarter Examinations.

Last day of summer quarter classes.

No classes. EE department is closed.

System-on-Chips (SoCs) are used everywhere - starting from simple IoT devices to complex safety-critical systems. Reusable hardware Intellectual Property (IP) based SoC design has emerged as a pervasive design practice in the industry to dramatically reduce SoC design and verification cost while meeting aggressive time-to-market constraints. Growing reliance on these pre-verified hardware IPs, often gathered from untrusted third-party vendors, severely affects the security and trustworthiness of computing platforms. An important emerging concern with the hardware IPs acquired from external sources is that they may come with deliberate malicious implants to incorporate undesired functionality, undocumented test/debug interface working as hidden backdoor, or other integrity issues. Evaluation of integrity and trustworthiness of third-party IPs is a critical emerging need for SoC designs. In this talk, I will survey ongoing security and trust validation efforts. Specifically, I will discuss how the complementary abilities of simulation-based validation, formal verification as well as side channel analysis can be effectively utilized for comprehensive SoC security and trust validation.

High reliability and availability is a requirement for most technical systems. Reliability and availability assurance methods based on probabilistic models is the topic addressed in this talk. Non-state-space solution methods are often used to solve models based on reliability block diagrams, fault trees and reliability graphs. Relatively efficient algorithms are known to handle systems with hundreds of components and have been implemented in many software packages. Nevertheless, many practical problems cannot be handled by such algorithms. Bounding algorithms are then used in such cases as was done for a major subsystem of Boeing 787. Non-state-space methods derive their efficiency from the independence assumption that is often violated in practice. State space methods based on Markov chains, stochastic Petri nets, semi-Markov and Markov regenerative processes can be used to model various kinds of dependencies among system components. However, the resulting state space explosion severely restricts the size of the problem that can be solved. Hierarchical and fixed-point iterative methods provide a scalable alternative that combines the strengths of state space and non-state-space methods and have been extensively used to solve real-life problems. We will take a journey through these model types via interesting real-world examples chosen from IBM, Cisco, Sun Microsystems, and Boeing. These methods and applications are fully described in a recently completed book: Reliability and Availability Engineering: Modeling, Analysis and Applications, Cambridge University Press, 2017.