Technical Program

Abstracts

"Honest Signals", Alex Pentland

We are in the midst of an explosion of information about people and their behavior, but most of it is noise. Reality Mining sifts through this noise to discover the `honest signals' hidden within: subtle patterns that reliably reveal our relationships with other people, and accurately predict our future behavior. Honest signals offer an unmatched window into our financial, cultural, and organizational health. By understanding these subtle patterns we can better understand ourselves, and begin to create true collective intelligences.

"Using Sensors to Make Sense of People", Tanzeem Choudhury

Systems that can capture and recognize human behaviors in unconstrained real world settings have the potential to dramatically impact research in smart environments, social network analysis, novel user interfaces, surveillance, and health care. There is a growing demand for behavioral monitoring systems in a wide range of health and wellness applications, such as fitness monitoring, elder-care support, long-term preventative and chronic care, and assisting those with cognitive disorders. In this talk, I will present our work in developing mobile sensing and machine learning techniques for reasoning about human activities, interactions, and social networks in everyday environments. I will also highlight how we tackle some of the research challenges in building activity-aware systems.

"Brains, and their applications", Richard Granger

Brains consist of components that are low-precision (2-3 bit), sparsely connected (p < 0.01), and very slow (milliseconds), yet they outperform competing approaches on a range of fundamental applications such as visual and auditory recognition. Brain circuits are circuits, albeit with unusual engineering designs, and they are becoming understood computationally. This talk will characterize typical brain circuit architectures and their operating rules; show examples of how such circuits give rise to novel algorithms for learning; and illustrate instances of these methods used in robot applications.

"Digital Tampering and Forensics", Hany Farid

Photography lost its innocence many years ago. Shortly after the first commercially available camera was introduced, photographs were being manipulated and altered. With the advent of high-resolution digital cameras, powerful personal computers and sophisticated photo-editing software, the manipulation of digital images is becoming more common. We are seeing the impact of these technologies in nearly every corner of our lives. As the technology that allows for digital media to be manipulated and distorted is developing at break-neck speeds, our understanding of the technological, ethical, and legal implications is lagging behind. I will discuss some of these issues and describe computational techniques which we have developed for detecting tampering in digital media.

"Printer Profiling for Forensics and Ballistics", Eric Kee

We describe a technique for authenticating printed and scanned text documents. This technique works by modeling the degradation in a document caused by printing. The resulting printer profile is then used to detect inconsistencies across a document, and for ballistic purposes that of linking a document to a printer.

"Flexible Object Manipulation: Tying Knots and Folding Laundry", Matthew Bell

Manipulation of flexible objects like string and cloth poses a challenge to robotic and human manipulation due to the difficulty of regrasping a flexible object. Without sensing, it becomes nearly impossible to guess where a grasp should go. However, we can try to simplify the problem in several ways. For tying knots in string, we can use a fixture to continually grasp the string during the entire tying process. We have developed fixtures for overhand and square knots, and can use a slightly modified overhand fixture to tie the knot around an object. Additionally, we are exploring automated methods of designing these fixtures, as well as a new class of fixture that works reliably with string as well as wire. In the area of cloth manipulation we have explored laundry folding methods that require a minimal number of grasps. We are also working on defining the number of fingers that are needed to stably grasp a piece of cloth.