The University of Arizona
banner image
Raquel Torres Peralta

Raquel Torres Peralta

Department of Computer Science
University of Arizona

Gould-Simpson Building
1040 E. 4th Street
Tucson, AZ 85721

rtorres at cs dot arizona dot edu

raquel.torresperalta at gmail dot com


Computer Science, PhD. I graduated on September, 2012. My advisor at the U of A was Kobus Barnard. I worked with Ian Fasel, Paul Cohen and Clay Morrison in different projects. Currently I am a profesor at the University of Sonora, Mexico.

Areas of specialization:

Artificial Intelligence, Machine Learning, Data Mining, Human-Robot Interaction, Computer-Human Interaction and Social Networks


My work has been focused on  gesture and user recognition (on Multitouch devices), time-sequence analysis, Human-Robot Interaction and Data Mining in social networks. I was part of the University of Arizona MindsEye team, working on human-behavior recognition and social-interaction detection and description. I am interested Machine Learning applications and Data Mining.  

Projects / Experiments

I have worked in the next projects and experiments during the my PhD in collaboration with other students and Professors:

User recognition on multitouch devices. Multi-touch tablets allow users to interact with computers through natural gestures over a surface with the fingers, allowing them to manipulate digital objects directly by touch. One advantage of these devices is that they offer a collaborative space where several users can work on a task at the same time, for example in a meeting room or emergency strategy center. In the multi-user case, it can be useful to be able to recognize individuals at interaction time. In this project, we achieve user recognition and authentication for touch devices, based on shape of gesture and velocity. The samples were collected under real-world conditions in a mining business office, using a conventional multi-touch tablet, with no high-precision devices.

Charlie the Robot In this human-Robot interaction experiment, the participants were asked to teach Charlie, a pirate robot, how to find a threasure using text and free speech. Teachers were told they were interacting directly with the robot, ignoring there was a person manipulating it behind the scene, following a Wizard of Oz (WOZ) paradigm. The goal was to understand how humans teach the robot and what do they expect from the robot during the session.

The UAV Experiment In this exploratory WOZ experiment, participants were asked to teach a simple task to a Unmanned Aereal Vehicle (UAV) using different modes of interaction (procedures, teaching by example, feedback and testing) using an interface to construct commands through a menu. This experiment revealed the preferences of teachers when using different methods while teaching the virtual student. An important contribution was the discover of three different types of teachers and the constant use of explicit teaching.

MindsEye The Defense Advanced Research Projects Agency has contracted with 12 research teams to develop fundamental machine-based visual intelligence and three teams to develop system integration concepts. “Ground surveillance is a mission normally performed by human assets, including Army scouts and Marine Corps Force Recon,” DARPA said. “Military leaders would like to shift this mission to unmanned systems, removing troops from harm’s way, but unmanned systems lack a capability that currently exists only in humans: visual intelligence. The Defense Advanced Research Projects Agency (DARPA) is addressing this problem with Mind’s Eye, a program aimed at developing a visual intelligence capability for unmanned systems. From:Kurzweil

Activity Recognition experiments The goal is to accomplish activity recognition from video input. I used the ICPR video database available online at ( The videos were processed to obtain the pose estimation using Deva Ramanan's code. The pose estimation is presented in form of boxes representing the different parts of the body. These boxes are processed to obtain a skeleton. To improve the quality of the pose across time, the whole sequence is smoothed using a method to manage outliers improving results significantly. In this video , the red skeleton shows the smoothed version of the original (in colors). We have used SVMs to identify the activities on video. The experiments have involved different features as head-fluents motion of both individuals, motion angle of each key-point as head, hand, knee, elbow, and feet.

Early Cancer Detection using thermography (In collaboration with the Physics department at the University of Sonora - DIFUS) The project explores the possibility of using thermography as an alternative to expensive and more invasive techniques to detect breast cancer on early stages analyzing the patterns on changes of temperatures across time on thermography files from voluntaries who have been diagnosed with breast cancer. The research is still in process.

Sentimental Analysis on Twitter Language processing on tweets to detect a positive, negative or neutral sentiment about a subject or person in one of the most influential networks today. The task represents a challenge in countries as Mexico where the meaning of a sentence depends on context and cultural background, and a word could be interpreted as a good or bad adjetive depending on timing.


Hagamos ciencia: ciencia viva, ciencia portátil, ciencia interactiva. Universidad, ciencia y cultura: evocaciones para un saber colectivo. Luz María Ortega Villa y Carlos Enrique Orozco Martínez, Coord. Universidad Autónoma de Baja California, 2014. ISBN 9786076071908

Learning a Policy for Gesture-Based Active Multi-touch Authentication. R. T. Peralta, A. Rebguns, I. Fasel, and K. Barnard. Human Aspects of Information Security, Privacy, and Trust. 2013. ISSN 0302-9743

Challenges to Decoding the Intention Behind Natural Instruction. Raquel Torres Peralta, Tasneem Kaochar, Clayton T. Morrison, Thomas J. Walsh, Ian R. Fasel and Paul R. Cohen. 20th IEEE International Symposium on Robot and Human Interactive Communication (Ro-Man 2011).

Challenges to Decoding the Intention Behind Natural Instruction (Extended abstract). Raquel Torres Peralta, Tasneem Kaochar, Clayton T. Morrison, Thomas J. Walsh, Ian R. Fasel and Paul R. Cohen. Raquel Torres Peralta, Tasneem Kaochar, Clayton T. Morrison, Ian R. Fasel, Thomas J. Walsh, Paul R. Cohen. IJCAI 2011 Workshop on Agents Learning Interactively from Human Teachers (ALIHT).Best presentation award

Human Natural Instruction of a Simulated Electronic Student. Tasneem Kaochar, Raquel Torres Peralta, Clayton T. Morrison, Thomas J. Walsh, Ian R. Fasel, Sumin Beyon, Anh Tran, Jeremy Wright and Paul R. Cohen. AAAI Spring Symposium: Help Me Help You (2011).

Towards An Understanding of How Humans Teach Robots. Tasneem Kaochar, Raquel Torres Peralta, Clayton T. Morrison, Ian R. Fasel, Thomas J. Walsh, Paul R. Cohen. User Modeling, Adaptation and Personalization (UMAP 2011).

Interactive Gesture-based Authentication on Tabletop Devices. Raquel Torres Peralta, Antons Regbuns, Ian R. Fasel. Technical Report, University of Arizona, Computer Science Department (2011).

Other activities

Columnist at the news site My collaborations are mainly about Science and Technology.

Co-Founder of the site K-Bits, news about Science and Technology.

Member of the RED ITIAM , Innovación y Trabajo en la Industria Automotriz Mexicana (ITIAM NETWORK, Innovation and Work for the Automotive Mexican Industry).

Member of CONAL , Consejo Nacional de Líderes Sociales de México (National Council of Social Leaders of Mexico).


Tópicos de matemáticas discretas (Discrete Mathematics) - University of Sonora

Introducción a la Ingeniería de Sistemas de Información (Introduction to Information Systems) - University of Sonora

Minería de Datos (Data Mining) - University of Sonora, TA University of Arizona

Robotica (Robotics) - University of Sonora

Negocios electrónicos. Modulo de procesamiento de datos (e-Bussiness, data processing module) - University of Sonora


Teaching Robots. Raquel Torres Peralta, Presentation at the Seminar of Computer Science, Cal Poly Pomona.

Challenges to Decoding the Intention Behind Natural Instruction. Presentation at ROMAN, 2011.

Algoritmos, introducción y ejercicios básicos. Raquel Torres Peralta, UNISON.

Proposiciones Logicas y Tablas de Verdad. Raquel Torres Peralta, Clase de Topicos de Matematicas Discretas, UNISON.


PhD in Computer Science, University of Arizona.

Masters in Computer Science, University of Arizona.

Licenciatura en Informática, instituto Tecnológico de Hermosillo.