Lisa Anthony
-- Researcher in Human-Computer Interaction --


Multimodal Stress Detection

My main project is exploring ways in which a system can detect, and then adapt to, user "stress" from variations in their input in multimodal interfaces. My collaborator is Andrew Sears at UMBC. We are currently exploring what types of sensors, and what features from data collected by those sensors, will be most indicative of stress, as well as defining what exactly we mean by "stress." We are drawing on work in affective computing, accessible computing, multimodal interfaces and interaction, machine learning, and human-computer interaction.

$N Multistroke Pen Gesture Recognizer

With my collaborator at UW, Jacob Wobbrock, I have been working on an extension to the $1 unistroke gesture recognition algorithm. The goal of $N, like $1 before it, is to implement a simple, fast, reasonable accuracy recognizer that can be easily incorporated into rapid prototyping processes to give application designers the power of trying out different input methods without having to spend a lot of effort getting the recognition up and running. You can try out a JavaScript implementation here, and check out the detailed pseudocode listing here. $N has some limitations, of course, but in our preliminary experiments, it has performed comparably to well-known recognizers, with much lower start-up costs. See our Graphics Interface 2010 paper here.

Project Homepage


User-Centered Interfaces for Military End Users

My work with the User-Centered Interfaces group at Lockheed Martin Advanced Technology Laboratories focused on applying advanced user interface technologies such as multimodal interaction and context-sensitive systems to the needs and requirements of the military end user. We deal with users at all echelons, in all domains, for many different purposes. Some examples of the types of projects we have worked on are described here. I was involved in work on I2W, SLICE, DARPA's ULTRA-Vis program, as well as various internal research projects.

Multimodal Algebra Equation Solving

My PhD thesis topic focused on the integration of handwriting-based interfaces with intelligent tutoring systems for mathematics. Traditional lessons in Cognitive Tutors on how to solve algebra equations require students to interact with an interface very different than that of solving equations on paper. Keyboard-and-mouse-based interfaces impose extraneous cognitive load on students during problem-solving and distract them from the math. A natural interaction experience using the handwriting modality can directly support the two-dimensional math notations such as fractions and exponents that more constrained, traditional interfaces do not. Yet, state-of-the-art handwriting recognition accuracy may not be good enough for students to use to solve problems, distracting them with the need to fix recognition errors. With my co-advisors, Ken Koedinger at CMU and Jie Yang now at NSF, I designed, developed and evaluated a proof-of-concept handwriting-based interface for an algebra tutoring system. On the user interaction and learning side, we found that students are able to solve more problems in the same amount of time, while achieving similar learning gains on the post-test, when using handwriting vs typing. On the technology and algorithms side, we found that the use of problem-solving context significantly improved handwriting recognition for the algebra domain.

Project Homepage

Rapid-Serial-Visual-Processing Interfaces for Video Search

During my summer internship at FXPAL in California, I was part of a team that designed and developed a Rapid-Serial-Visual-Processing (RSVP) interface for search and retrieval of video clips. The proof-of-concept application for the work was NIST's TRECVID (Digital Video Retrieval) competition. This interface and interaction paradigm is based on characteristics of human perception and visual search strategies. The RSVP interface presents imagery rapidly and automatically to a user with a specific information need, and provides fine-grain controls that the user manually activates when search target(s) are near. A CSCW 2008 paper by the FXPAL team can be found here.

 Active Learning in Problem Solving (ALPS)

Students working in Cognitive Tutors (originally built by Carnegie Mellon's PACT Center) receive dynamic feedback and help upon request while they are problem-solving, but they are not able to ask conceptual questions that go beyond the problem currently being solved. ALPS is an approach designed to provide students with such a mechanism, by integrating Synthetic Interview technology with the existing Cognitive Tutor. Students can type in any question in their own words, while working in the Tutor, and they receive a pre-recorded video clip in response. Because they are pre-recorded, the designers must get a sense of the types of questions to expect to receive, and how best to categorize and answer them. Preliminary results from this line of work have shown that students ask questions in such a system with heavy emphasis on performance goals. A pdf of the Intelligent Tutoring Systems conference paper about this work is here. An open future area of research is how to encourage students to ask deeper questions and shift their goals more toward learning. Until 2004, I helped collect data on student question-asking patterns during tutoring and types of instruction that encourage more active question-asking. The current group's page is here.

 Evolving Board Evaluation Functions via Genetic Programming (Master's Thesis)

For my Master's thesis at Drexel University, I applied genetic programming techniques to the evolution of a strategy for evaluating potential moves in a one-step lookahead intelligent agent heuristic for a complex strategy-based game. The game I used is Acquire, whose object is to amass wealth while investing stock in hotel chains and effecting mergers of these chains as they grow. Genetic programming was used to evolve the board evaluation functions used by agent players of the game. The analysis of game interactions is recognized as a valid analogy to common real-world problems, which often present difficulty in designing algorithms to solve them. Genetic programming, as a branch of evolutionary computation, provides advantages over traditional algorithms in solving these complex real-world problems in speed, robustness and flexibility. The thesis contributed to work in artificial intelligence in providing computer systems with the tools they need to learn how to operate within a domain, given only the basic building blocks. A pdf version of the thesis document is here.

 Conceptual Understanding and Prototyping (CUP)

My main undergraduate research project was CUP, a 3D conceptual modeling system for assembly design. At the time, many tools existed to support computer-aided design in engineering disciplines during mature stages of the design process only. The conceptual design phase, which focuses on not the geometry but the structure, behavior and function of an artifact, however, can often determine a large percentage of the overall cost of the product, as many fundamental decisions are made during this stage. CAD tools available often forced designers to make detailed geometry or layout decisions without allowing a means for markup of an artifact's structure, function and behavior. CUP was designed as a tool to allow engineers to capture the design intent inherent in the conceptual design process. The definitive CUP journal paper is here. A downloadable Java version of the software is still available here. CUP has since been extended into a secure, multi-user collaborative conceptual design environment called MUG, and is an ongoing project at the Geometric and Intelligent Computing Laboratory at Drexel University.

 Acquire Game-Playing System

This project was a group project for the Software Engineering Workshop capstone course as part of the computer science degree at Drexel University. My teammates were Luiza da Silva, Mike Czajkowski, and Xuan Thuy Tran. We built a distributed system of intelligent agents playing the game of Acquire with utility theory-based strategies, starting from the ground up with a software engineering requirements document and design document and then building the system. This project is no longer maintained, but you can see our project webpage created for the class here.


 Robot Lab

In my senior year at Drexel University, I took a robot building laboratory class. For the final project, I investigated the accuracy of two of the default methods for using the sonar sensors that we used with our hardware kits. I did this via a simple experimental design wherein I recorded the sonar sensor's readings of distance at certain known distances and orientations. This project is no longer maintained, but you can see the webpage I created for it for the class here.


 Introduction to HCI (Drexel)

My very first HCI project -- a group project for the final grade in a course taught by Dr. Tom Hewett at Drexel University. We reported a case study of three engineers using a paper mock-up of an interface for searching design repositories for past artifacts based on their structure, function and behavior annotations, in the conceptual design phase of building a new artifact. You can read the paper that was the result of this project, including the case studies, here.


 Software Design -- KWIC (Keyword in Context)

For a software engineering course called Software Design, my team and I had to implement a simple "keyword in context" program using 3 different software design patterns -- implicit invocation, pipe-and-filter, and object-oriented. We were then supposed to compare the three processes and implementations. KWIC is a standard software engineering toy problem; it is an indexing system consisting of all "circular shifts" (repeatedly removing the first word and appending it to the end of a line) of a particular set of lines in alphabetical order. This project is no longer maintained, but you can see our project webpage created for the class here.
(c) 2004-2011 Lisa Anthony. Last revised 03/03/2011.