Thesis (Ph. D.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2006. / Includes bibliographical references (leaves 159-169). / As the study of complex interconnected networks becomes widespread across disciplines, modeling the large-scale behavior of these systems becomes both increasingly important and increasingly difficult. In particular, it is of tantamount importance to utilize available prior information about the system's structure when building data-driven models of complex behavior. This thesis provides a framework for building models that incorporate domain specific knowledge and glean information from unlabeled data points. I present a methodology to augment standard methods in statistical regression with priors. These priors might include how the output series should behave or the specifics of the functional form relating inputs to outputs. My approach is optimization driven: by formulating a concise set of goals and constraints, approximate models may be systematically derived. The resulting approximations are convex and thus have only global minima and can be solved efficiently. The functional relationships amongst data are given as sums of nonlinear kernels that are expressive enough to approximate any mapping. Depending on the specifics of the prior, different estimation algorithms can be derived, and relationships between various types of data can be discovered using surprisingly few examples. / (cont.) The utility of this approach is demonstrated through three exemplary embodiments. When the output is constrained to be discrete, a powerful set of algorithms for semi-supervised classification and segmentation result. When the output is constrained to follow Markovian dynamics, techniques for nonlinear dimensionality reduction and system identification are derived. Finally, when the output is constrained to be zero on a given set and non-zero everywhere else, a new algorithm for learning latent constraints in high-dimensional data is recovered. I apply the algorithms derived from this framework to a varied set of domains. The dissertation provides a new interpretation of the so-called Spectral Clustering algorithms for data segmentation and suggests how they may be improved. I demonstrate the tasks of tracking RFID tags from signal strength measurements, recovering the pose of rigid objects, deformable bodies, and articulated bodies from video sequences. Lastly, I discuss empirical methods to detect conserved quantities and learn constraints defining data sets. / by Benjamin Recht. / Ph.D.

Thesis (Ph.D.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 1999. / Includes bibliographical references (p. 246-258). / This thesis presents new theory and technology for the representation and recognition of complex, context-sensitive human actions in interactive spaces. To represent action and interaction a symbolic framework has been developed based on Roger Schank's conceptualizations, augmented by a mechanism to represent the temporal structure of the sub-actions based on Allen's interval algebra networks. To overcome the exponential nature of temporal constraint propagation in such networks, we have developed the PNF propagation algorithm based on the projection of IA-networks into simplified, 3-valued (past, now, future) constraint networks called PNF-networks. The PNF propagation algorithm has been applied to an action recognition vision system that handles actions composed of multiple, parallel threads of sub-actions, in situations that can not be efficiently dealt by the commonly used temporal representation schemes such as finite-state machines and HMMs. The PNF propagation algorithm is also the basis of interval scripts, a scripting paradigm for interactive systems that represents interaction as a set of temporal constraints between the individual components of the interaction. Unlike previously proposed non-procedural scripting methods, we use a strong temporal representation (allowing, for example, mutually exclusive actions) and perform control by propagating the temporal constraints in real-time. These concepts have been tested in the context of four projects involving story-driven interactive spaces. The action representation framework has been used in the Intelligent Studio project to enhance the control of automatic cameras in a TV studio. Interval scripts have been extensively employed in the development of "SingSong ", a short interactive performance that introduced the idea of live interaction with computer graphics characters; in "It/I", a full-length computer theater play; and in "It", an interactive art installation based on the play "It /I" that realizes our concept of immersive stages, that is, interactive spaces that can be used both by performers and public. / by Claudio Santos Pinhanez. / Ph.D.

Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2004. / Includes bibliographical references (p. 65-68). / Online personal advertisements have shed their stigma as matchmakers for the awkward to claim a prominent role in the social lives of millions of people. Web sites for online dating allow users to post lengthy personal ads, including text and photos; search the database of users for potential romantic partners; and contact other users through a private messaging system. This work begins with psychological and sociological perspectives on online dating and discusses the various types of online dating Web sites. Next, it presents an analysis of user behavior on one site in particular, which has more than 57,000 active users from the United States and Canada. A demographic description of the population is given, and then 250,000 messages exchanged by the active users over an eight-month period are analyzed. An examination of which characteristics are "bounding" finds that life course attributes such as marital status and whether one wants children are most likely to be the same across the two users in a dyadic interaction. To understand which characteristics are important to users in deciding whom to contact, regression models show the relative strength of a variety of attributes in predicting how many messages a user with those attributes will receive. By far the strongest predictor of messages received is the number of messages sent. For men, age, educational level, and self-rated physical attractiveness are the next most important qualities. For women, they are not being overweight, self-rated physical attractiveness, and having a photo. Finally, a discussion of the design implications of these findings and other design issues follow the results. / Andrew Rocco Tresolini Fiore. / S.M.

Thesis (Ph. D.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2006. / Includes bibliographical references (p. 106-108). / Whenever we use computers to interact with media, our experience is that of direct control, and the goal of our interactions is either artifact-production (the editor paradigm) or passive exploration (the browser paradigm). This thesis proposes an alternative: a model of media interaction based on the ideas of non-idiomatic improvisation that encourages active exploration of media and its structures. We argue that in order to facilitate this kind of exploration, (1) computational tools must actively participate in the creative process and (2) the interaction framework must allow structural exploration of media. This leads to our main claim: improvisation should be considered a valid and appropriate paradigm for media interaction. To this extent, we present a Cognitive Model of Improvisational action (CMIA) that integrates element-centric and process-centric (structural) modes of control into a single framework for media exploration. This model allows participants to switch their attention between compositional elements and structural processes. / (cont.) The model is argued to be particularly powerful in leading us to novel spaces for media creation and consumption. We follow by presenting the Emonic Environment (Implementation), an interactive system built on the principles of CMIA. We conclude by describing two studies (Scenarios a Experiments) that analyze the ways in which Emonic Environment affects how people interact and think about their interactions with digital media. These studies illustrate the potential of CMIA as a paradigm for interaction between humans and machines. / by Paul Nemirovsky. / Ph.D.

Thesis (M.S.)--Massachusetts Institute of Technology, Program in Media Arts & Sciences, 1996. / Includes bibliographical references (leaves 60-65). / by David José Tamés. / M.S.

Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2008. / Page 96 blank. / Includes bibliographical references (p. 81-82). / This thesis introduces 'Quickies', an attempt to bring one of the most useful inventions of the 20th century into the digital age: the ubiquitous sticky notes. Sticky notes help us manage our to-do lists, tag our objects and documents and capture short reminders or information that we may need in the near future. 'Quickies' enrich the experience of using sticky notes by allowing them to be tracked and managed more effectively. Quickies are sticky notes that have a digital duplicate and the ability to remind us about the task we ought to perform or to provide us at the right time with the information we captured in the past. The thesis explores how the use of Artificial Intelligence (AI), Natural Language Processing (NLP), RFID, and ink recognition technologies can make it possible to create intelligent sticky notes that can be searched, located, can send reminders and messages, and more broadly, can help us to seamlessly connect our physical and digital information worlds. / by Pranav Mistry. / S.M.

Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2012. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 83-86). / Lower-extremity amputees face potentially serious post-operative complications, including increased risk of further amputations, excessive stress on both limbs, and discomfort at the stump/socket interface. State of the art; passive prostheses have improved many negative consequences associated with lower-limb loss, but we believe the limit of uninformed elastic prostheses has been reached. Further strides require a more biomimetic approach. Through integration of "smart" technology (sensors and actuators), a new phase of bionic lower-limb prostheses is upon us, which enables prosthetic devices to more closely mimic biological behavior by generating human-like responses and power outputs. The closer we come to natural biology, gait abnormalities in amputees will decline. This project compares the first bionic ankle prosthesis to commonly used passive prostheses to determine how more biomimetic adaptability and work generation in the prosthetic joint affects discomfort and joint stress. We have put forth several metrics to describe discomfort (elements of shock absorption, pressure distribution, etc.) and will conduct level-ground walking tests with three unilateral amputee subjects using both passive and power devices. We hope to make a case for the pursuit of more biomimetic designs for rehabilitative devices, by showing a positive effect on "comfort" and a restoration of normal gait dynamics when using a bionic ankle prosthesis. / by David Allen Hill. / S.M.

Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2012. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 85-89). / This thesis aims at discovering algorithms to recover the geometry of hidden objects from tertiary diffuse scattering, given time of flight information. We focus on using ultra high speed capture of photons to accurately determine information about distance light travelled and using it to infer hidden geometry. We aim at investigating issues such as the feasibility, uniqueness(in solution domain) and invertibility of this problem. We also aim at formulating the forward and inverse theory of secondary and tertiary diffuse scattering using ideas from tomography. We aim at developing tomography based approaches and sparsity based methods to recover 3D shapes of objects "around the corner". We analyze multi-bounce propagation of light in an unknown hidden volume and demonstrate that the reflected light contains sufficient information to recover the 3D structure of the hidden scene. We formulate the forward and inverse theory of secondary and tertiary scattering reflection using ideas from energy front propagation and tomography. We show that using careful choice of approximations, such as Fresnel approximation, greatly simplifies this problem and the inversion can be achieved via a backpropagation process. We provide a theoretical analysis of the invertibility, uniqueness and choices of space-time-angle dimensions using synthetic examples. We show that a 2D streak camera can be used to discover and reconstruct hidden geometry. Using a 1D high speed time of flight camera, we show that our method can be used recover 3D shapes of objects "around the corner". / by Otkrist Gupta. / S.M.

Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2005. / Includes bibliographical references (p. 244-251). / The sense of touch is one of the most important senses of the human body. This thesis describes the biologically inspired design of "sensitive skins" for two different robotic platforms: Leonardo, a high degree-of-freedom, sociable robot and the Huggable, a portable therapeutic robotic companion for relational, affective touch. The first step in the design of the "sensitive skin" for Leonardo: a set of hands featuring 40 force-sensing resistors (FSRs) and embedded processing was created. Somatosensory inspired algorithms for calculating the location, direction of motion, and orientation with a set of these sensors forms the first stage in the design of a "Virtual Somatosensory Cortex." A multi-modal (temperature, electric field sensors, and Quantum Tunneling Composite (QTC) based force sensors) three dimensional sensor array was created as the first step in the design of the "sensitive skin" for the Huggable. A soft silicone skin was placed over this array. Preliminary results using neural networks show that the affective content of touch can be determined. This work was sponsored in part by the NSF Center for Bits and Atoms Contract No.CCR-0122419, a Microsoft iCampus grant, and the MIT Media Lab Things That Think and Digital Life Consortia. / by Walter Dan Stiehl. / S.M.

Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2007. / Includes bibliographical references (p. 89). / This thesis explores how natural bodily movements can be translated into a control interface for vehicles. Focusing on the car, our goal is to increase human performance and wellbeing while eliminating the traditionally antagonistic relationship between comfort and freedom of movement vs. support, safety and sensing the car. We will discuss seating, traditional controls, their origins, evolution, and their implications in the context of today's cars. Based on the physical demands of the vehicle environment, and on positive body experiences from sports and other concepts of movement, we will then explore how we could re-think the function, self-image, and presentation of the human body in the context of cars. We will develop a seat prototype, which will encourage beneficial body sensations and - motions, taking into account the shapes, textures, and emotional significance of touch and movement in and by itself, and in the car environment. The core of our concept will focus on natural movements of the lower back and hips, as experienced when walking or skiing. Building on the exoskeleton-like "Athlete Seat," which blurs the boundaries between wearing and sitting in, we will develop the core prototype out towards the upper body and limbs. / (cont.) We will develop a second prototype, which will have pelvic movements in the frontal plane as done when walking, bicycling, or dancing, as the basis of its concept. This prototype will be connected to a car simulator to investigate if good vehicle control can be achieved with our method. In a second stage, we will systematically evaluate the car control, wellbeing, and fun aspects in a user study. Our modular design will be usable in parts and adaptable to various uses, in vehicles, for entertainment, exercise, wellbeing, and medical purposes, improving physical condition and the way we relate to our bodies. / by Patrik A. Künzler. / S.M.