Colloquium Archive

Fire Sales And Search

John Mamer, UCLA Anderson School of Management, Los Angeles, CA


We study a model asset sales via search with semi-rational buyers. A seller has a finite (or infinite) number of items to sell; buyers arrive according to a Poisson process. An arriving buyer presents an offer to the seller. Buyers' values for the asset are modeled as independent draws from a probability distribution. The seller must decide whether or not to accept the offer or to reject it and wait for a better one. The seller pays a cost per unit time to continue searching, and has a discount rate. Each potential buyer knows the price of the last sale, and offers the minimum of his reservation value and the price of the last sale. As a result, the seller faces a falling offer distribution, each sales price setting the maximum offer for subsequent sales. Our model has application to distributor/manufacturer negotiations and pricing of technology products.

The *Puf* Makes The Difference

Sid Paral and Mandel Yu, Verayo, San Jose, CA


Almost daily, we hear about outrageous cases of product counterfeiting; we all experience increased demands for identification of persons and computer data. Traditional concepts of trust, based on scarcity and secrets, are eroded by perfect digital copies, both in software and hardware. Physical Unclonable Functions (PUFs) offer a solution: they reference objects to measurable qualities outside the creator’s (or would-be counterfeiters) control, with the help of naturally occurring randomness. In our team, we focus on silicon chips built in a standard manufacturing process, and in particular, on their differential timing performance, rooted in random manufacturing process variations - still within normal circuit specifications. To create reliable and repeatable PUF, the design must emphasize uniqueness while suppressing systematic patterns, digitizing noise, and environmental influences. As a practical example, we will explain the operation of a basic, challengeable, multiplexer-array PUF, explore its response characteristics, choose an operational threshold, and determine its nominal error rates.

Hosting And Managed Services For Ecommerce

Dario Zea, MarketLive, Petaluma, CA


Successful ecommerce sites require a reliable and robust infrastructure that scales horizontally as user demand increases. This infrastructure needs to be able to recover quickly from downtime potentially caused by multiple sources such as the main application engine, resource utilization across environments or even external denial of service attacks. For this talk we will go over what it takes to manage networks and systems for high availability, tools used to oversee performance and capacity, and technologies used to deploy a production environment.

Non-Invasive Malicious Javascript Detection

EJ Jung, Computer Science, University of San Francisco, San Francisco, CA


Malicious JavaScript detection is challenging in several ways. First, it is usually invisible to users, so it cannot be avoided by users’ vigilance. Second, it is often heavily obfuscated to bypass signature-based detection mechanisms and term-based classifiers such as the Bayesian classifier often used for spam detection. We have combined a web crawler for targeted collection of malicious JavaScript, a de-obfuscator that derives URLs from malicious JavaScript, and multiple classifiers for a comprehensive detection framework. Among tested, Support-Vector Machine showed the best performance. Given our test set from well-known malicious JavaScript repositories and high-traffic websites, our framework detects around 90% of malicious JavaScript and 99.9% of good scripts correctly. These results are published in Malware 2009, ACM DSMM '09, and ACM CCS '09 in poster session. This detection rate is already useful in real-time detection, especially when blacklists are not up to date and static and dynamic analysis imposes too much delay for web surfing. To improve the detection even further, we are currently building more detailed-blacklists using our crawlers.

Cpu Vs. Gpu: When Worlds Collide

Jayanth Gummaraju, Advanced Micro Devices, Sunnyvale, CA


CPUs and graphics processing units (GPUs) have been evolving independently, attaining preeminence in their own domains. Recently, they've been encroaching on each others' turf, setting the stage for drastic changes in software design and applications. This talk will review the current trends in CPU and GPU evolution, and suggest likely challenges and opportunities for the exciting times that lie ahead.