Auctions for selling advertisement space have been the main mechanism used to monetize Internet services. These auctions give rise to a number of interesting challenges not traditionally considered by auction theory. Search advertising gives rise of an enormous number of auctions running simultaneously, necessitating the use of extremely simple mechanisms. Traditional auction theory tells us how to design optimal auctions (maximizing welfare or revenue), but typically results in designs that are too complex for the web environment, where it is essential for auctions to have extremely clear and simple design.
Over the last 10+ years we have developed a good understanding of many games naturally arising in the context of Internet or web services from the perspective of the resulting social welfare, including a good understanding of games modeling selfish traffic routing, service location, bandwidth sharing among others. In this talk we will consider auctions from this perspective in various settings including the commonly used auction format Generalized Second Price.
Auctions for selling advertisement space have been the main mechanism used to monetize Internet services. These auctions give rise to a number of interesting challenges not traditionally considered by auction theory. Search advertising gives rise of an enormous number of auctions running simultaneously, necessitating the use of extremely simple mechanisms. Traditional auction theory tells us how to design optimal auctions (maximizing welfare or revenue), but typically results in designs that are too complex for the web environment, where it is essential for auctions to have extremely clear and simple design.
Over the last 10+ years we have developed a good understanding of many games naturally arising in the context of Internet or web services from the perspective of the resulting social welfare, including a good understanding of games modeling selfish traffic routing, service location, bandwidth sharing among others. In this talk we will consider auctions from this perspective in various settings including the commonly used auction format Generalized Second Price.
Typical models of strategic interactions in computer science use simultaneous move games. However, in applications simultaneity is often hard or impossible to achieve. In this talk we study the robustness of the Nash Equilibrium when the assumption of simultaneity is dropped. In some classes of games sequential action significantly improves the quality of the predicted solution, resulting in much more natural and better quality prediction. In the context of auctions, sequential implementation gives players interesting strategic opportunities and introduces externalities in the auction game. We study such games in an attempt to understand the role of simultaneity in simple auction formats (such as simultaneous first or second price auctions).
Coffee served at 3:30pm in WID room 3280, Presentation starts at 4:00pm
Anyone without WID access can use the special event elevator on the WID 1st floor (near ALDO Café) to access the 3rd floor teaching lab
Speaker:Jean-Luc Thiffeault
Associate Professor, Department of Mathematics, UW-Madison
Presentation Title:
Topological optimization
Abstract:
Topological chaos is a type of chaotic behavior that is 'forced' by the motion of obstacles in some domain.I will review two topological approaches, with applications in particular to stirring and mixing in fluid dynamics.The first approach involves constructing systems such that the fluid motion is topologically complex, usually by imposing a specific motion of rods.I will then discuss optimization strategies that can be implemented.The second approach is diagnostic, where flow characteristics are deduced from observations of periodic or random orbits and their topological properties.
Speaker: Elizabeth S. Burnside, MD PhD Department of Radiology; Department of Biostatistics and Medical Informatics
University of Wisconsin School of Medicine and Public Health
Using Multi-relational Data and Machine Learning to Improve Breast Cancer Diagnosis Tuesday
4:00 pm
Biotechnology Center Auditorium, 425 Henry Mall
Abstract:
In the new era of "-omic"-based research, many scientists have shifted from the study of the individual parts of a system to the system itself. This new paradigm focuses on a comprehensive collection of a fundamental data type that can provide a platform for a myriad of research directions on a given level ranging from the subcellular to the population. However, developing methodologies that integrate these rich data sources to inform and improve healthcare decisions on the patient level remains challenging. Our team of physicians, computer scientists, and industrial engineers at the University of Wisconsin has collaborated for the last decade to develop methods to improve breast cancer diagnostic decision-making using inductive logic programming, statistical relational learning, advice-based-learning, and other cutting-edge machine learning techniques. Our algorithms are designed to utilize the ever-expanding, multi-relational data that predicts breast cancer including: genetic, imaging, and epidemiologic risk factors. This talk will present an overview of our research programs and provide a vision of the future of computational methods in the domain of breast cancer risk prediction.
Given the current trends in computing, it may be a good time to
rethink coherent memory in multicores. The extremes of computing where
we find multicores, like mobile devices and datacenters, have
drastically different characteristics than the discrete
multiprocessors where coherence emerged. Also, as architects pursue
deeper integration of accelerators like GPUs on die, full on-chip
coherence may no longer be optimal.
In addition to rethinking coherence because of ecosystem changes, we
show that coherence prevents a key opportunity for software to exploit
on-chip caches. Software runtimes that require isolation (e.g.,
transactional memory, database logs) usually make copies of data that
they modify -- copies which, ironically, may already exist in cache.
Because coherence hides caches from software, applications are forced
to make expensive and redundant copies, and consequently often require
hardware acceleration for good performance.
In this talk, we propose Acoherent Shared Memory (ASM), a new
abstraction that gives software control over how hardware manages
memory. ASM introduces the concept of acoherent memory that is neither
coherent (it allows multiple versions of the same address) nor
incoherent (hardware resolves differences when it matters). Acoherence
exposes private memory to threads, and uses a checkout/checkin
abstract to manage data movement between private and shared memory. We
show that ASM performs comparably to coherent systems for existing
workloads and can accelerate software data isolation up to 49%, paving
the way for fast software-only runtimes such as transactional memory.
Wisconsin Institute for Discovery (WID) Room 3280 (Anyone without WID access can use the special events elevator on WID 1st floor (near Aldo Cafe) to access 3rd floor teaching lab (rm 3280)
Coffee served at 3:30 in conjunction with WID Discovey. Non-building occupants meet at room 3280.
Sushmita Roy, Asst Professor, Biostatistics and Medical Informatics, UW-Madison
Title: Learning networks in biology: opportunities and challenges
A systems-level understanding of how living system function requires us to identify the parts of a system and the interactions between these parts. More importantly, the interactions of a system may be “condition-specific”, where a condition could represent a different environmental stress, a cell-type, a disease or different organisms. Advances in biotechnology are enabling us to identify the parts of a system, however, identifying the interactions remains a difficult problem. I will discuss the inference and analysis of these networks in the usual setting of learning a single network, and then in a more interesting setting of simultaneous learning of multiple networks. I will present the challenges in learning these networks, specifically, when we aim to do this at the genome-scale, with thousands of nodes. I will present some of our approaches, based on probabilistic graphical models, to address this problem that allows us to use search heuristics and incorporate prior knowledge into system to make the problem more tractable and the networks more biologically realistic.
Guests in a virtualized environment expect a contiguous,
isolated physical address space. To provide this, the hypervisor
translates guest physical addresses into host physical addresses, much
like virtual addresses are translated into physical addresses in an
operating system. There are two approaches to managing these
translations. With shadow page tables, the hypervisor traps on guest
page table modifications and updates the shadow page table, a list of
GVA->HPA translations. This requires no hardware support but involves
many costly VM exits. Nested page tables are an alternative where
both the guest and host page tables are hardware walked. This
eliminates the necessity of trapping into the hypervisor when the
guest page table is updated, but for n-level page tables can result in
O(n*2) accesses to translate each address. I will present a survey of
recent work to speed up address translation, including 2D translation
caching, addition of a nested TLB, and using a hashed page table for
the GPA->HPA translation.
Speaker:Andrea Thomaz, Georgia Institute of Technology
Cookies at 3:30 pm
Abstract: In this talk I present recent work from the Socially Intelligent Machines Lab at Georgia Tech. One of the focuses of our lab is on Socially Guided Machine Learning, building robot systems that can learn from everyday human teachers. We look at standard Machine Learning interactions and redesign interfaces and algorithms to support the collection of learning input from naive humans. This talk covers results on high-level task goal learning, low-level skill learning, and active learning interactions using several humanoid robot platforms.
Bio: Andrea L. Thomaz is an Assistant Professor of Interactive Computing at the Georgia Institute of Technology. She directs the Socially Intelligent Machines lab, which is affiliated with the Robotics and Intelligent Machines (RIM) Center and with the Graphics Visualization and Usability (GVU) Center. She earned a B.S. in Electrical and Computer Engineering from the University of Texas at Austin in 1999, and Sc.M. and Ph.D. degrees from MIT in 2002 and 2006. Dr. Thomaz is published in the areas of Artificial Intelligence, Robotics, Human-Robot Interaction, and Human-Computer Interaction. She received an ONR Young Investigator Award in 2008, and an NSF CAREER award in 2010. Her work has been featured on the front page of the New York Times, and in 2009 she was named one of MIT TR35.
