570 Vincent Hall
Goldman Sachs: Career opportunities for quantitative individuals
Abstract: Goldman Sachs is a leading global investment banking, securities and investment management firm that provides a wide range of services worldwide to a substantial and diversified client base that includes corporations, financial institutions, governments and high net worth individuals. Goldman Sachs has long been a destination for newly minted MBAs, but with the increasing complexity of financial products, the firm has become a major recruiter of highly quantitative individuals, often with little or no business training, who fill roles in derivative pricing, risk management and portfolio optimization. For these positions, the firm is particularly interested in individuals with a background of study in mathematics, physics, engineering, computer science or another highly quantitative discipline as well as excellent problem solving abilities and strong programming skills.
In this talk, I will attempt to convey, from my experience, a sense what it is like to work in the financial industry, the opportunities available to quantitative individuals and the nature of some of the mathematical problems that one encounters.
Abstract: Hepatitis C affects between 4 and 5 million people in the United States with nearly 75% suffering chronic infection. Further, current estimates indicate 170 million people are infected, worldwide, with the Hepatitis C virus (HCV). Pegylated interferon and ribavirin combination therapy is currently the standard of care for treatment of HCV but is considered less than optimal. For example, sustained virologic response (SVR) for genotype 1 virus is observed in pivotal clinical trials for only 54% of treated patients. SVR is defined as undetectable HCV RNA in plasma on the order of 6 months after cessation of treatment. This combination therapy is also associated with a high incidence of significant side effects suffered over a treatment interval of at least six months. Identification of better treatment alternatives is the goal of a vigorous antiviral program at Merck.
In this talk we describe mathematical models explaining HCV infection and response to treatment. The model equations and resultant simulations were chiefly derived from the literature. Simulations have already provided support for the antiviral drug development program at Merck. Mathematical modeling gives evidence of target engagement and drug efficacy. Using accumulated simulation experience gained from basic research, preclinical data, and the literature, design optimizations for early clinical studies may be proposed. As internal modeling expertise is enhanced by experience gained in early development, this mathematical knowledge base may help optimize costly phase III trials by guiding key decisions such as dose selection and length of dosing.
Networked target tracking architectures
Abstract: The coordinated use of multiple distributed sensors by network communication has the potential to substantially improve estimates of target positions, features, and attributes. This improvement is primarily due to geometric diversity, complementary sensor information, and different coverage areas. Unfortunately, sensor networks are not without their challenges. In particular, there is a balance between network bandwidth constraints and the maintenance of a consistent picture of the scenario across the network participants. In this talk we will discuss pertinent issues including network communication schemes, track initiation and maintenance, and bias mitigation.
Abstract: All marketing organizations — including the successful ones — are under increased pressure to do more with less. Throughout the marketing and fulfillment delivery chain marketers face competing business goals, multiple marketing programs and constraints like channel capacity, budget and customer contact policies. Typically it is required to maximize/minimize an objective such as the return on investment while satisfying these marketing goals/constraints. In this talk we will discuss the formulation of this problem as a binary integer programming model and outline the computational challenges involved in solving the model to optimality. We will also present an overview of the methodology used by the SAS Marketing Optimization software followed by some computational results.
Abstract: In computer aided diagnosis (CAD) applications the goal is to detect structures of interest to physicians in medical images: e.g. to identify potentially malignant lesions in an image (mammography, lung CT, Colon CT, heart ultrasound, etc.). In an almost universal paradigm, this problem is addressed by a 5 stage system:
1. Segmentation to identify/extract the general area of interest; 2. Candidate generation which identifies suspicious unhealthy candidate regions of interest (ROI) from a medical image; 3. feature extraction that computes descriptive features for each candidate; 4. classification that differentiates candidates based on candidate feature vectors; 5. visual presentation of CAD findings to the radiologist in order for him to accept or reject the CAD findings.
For the fourth stage, many standard algorithms (such as support vector machines (SVM), back-propagation neural nets, kernel Fisher discriminants) have been used to learn classifiers for detecting malignant structures. However, these general-purpose learning methods either make implicit assumptions that are commonly violated in CAD applications, or cannot effectively address the difficulties arisen when learning a CAD system.
Non-IID Data Traditional learning methods almost universally assume that the training samples are independently drawn from an identical albeit unobservable underlying distribution (the IID assumption), which is often not the case in CAD systems. Due to spatial adjacency of the regions identified by a candidate generator, both the features and the class labels of several adjacent candidates are highly correlated.
In this talk we present two recent proposed machine learning algorithms that successfully takes into account the correlation among candidates to significantly improve classification performance.
Interaction of lipid bilayers with inorganic surfaces: experiments to modeling
Joint work withand .
Abstract: Lipid bilayers typically serve as the scaffold for trans membrane protein receptors. These membrane protein receptors are targets for atleast 50% of the drug molecules developed by pharmaceutical companies. Drug development and testing typically is carried out in an invitro environment with these lipid membranes deposited on some synthetic surface organic or inorganic. Here we present details of the interactions between lipid bilayers and inorganic surfaces. The approach adopted to develop an understanding of these interactions has been a combination of a multiscale modeling framework in conjunction with specific experimental effort to complement the modeling effort. The multiscale modeling effort involves modeling the membrane surface interactions in detail at an atomistic level to looking at macroscopic membrane dynamics on surfaces with specific topologies. The experimental effort involves a combination of Surface Force Apparatus (SFA) and Atomic Force Microscopic (AFM) measurements to generate these insights. The talk will focus specifically on the role of surface topology in modulating membrane surface interactions.
Challenges for numerical analysis in medical device industries
Abstract: This presentation will provide an overview of how numerical analyses are currently being used in the development and reliability investigations for implantable medical devices. Specific emphasis will be given on the current methods being used and additionally the barriers to the usefulness of these numerical predictions.