Rafael Sousa Silva

Ph.D. Student, MIRRORLab, Colorado School of Mines

rsousasilva@mines.edu

Table of Contents

Research Interest

My research focuses on the implementation and evaluation of forgetting models from cognitive psychology, namely decay and interference, as the main dynamics of robotic working memory. The way in which information leaves working memory buffers can influence robot natural language processes, such as referring expression generation. Thus, my research goal is to use these forgetting methods in order to enable robotic natural language that is human-like, accurate, natural, easy to understand, and computationally inexpensive.

Education

2021 - present Robotics Ph.D. Student, Colorado School of Mines, Golden, CO

TA Experience: Algorithms

Coursework: Human-Robot Interaction; Computer Vision; Advanced Machine Learning; Robot Ethics; Robot Mechanics: Kinematics, Dynamics, and Control; Human-Centered Robotics; and Social & Collaborative Computing.


2017 - 2021 B.A. Computer Science, Franklin & Marshall College, Lancaster, PA

2017 - 2021 B.A. Mathematics, Franklin & Marshall College, Lancaster, PA

Publications

Rafael Sousa Silva, Michelle Lieng, and Tom Williams. Forget About It: Entity-Level Working Memory Models for Referring Expression Generation in Robot Cognitive Architectures. In Proceedings of the Annual Meeting of the Cognitive Science Society, 2023. (UNDER REVIEW)


Rafael Sousa Silva and Tom Williams. Enabling Human-like Language-Capable Robots Through Working Memory Modeling. In Companion of the 2023 ACM/IEEE International Conference on Human-Robot Interaction (HRI '23 Companion), 2023.


Will Culpepper, Thomas A Bennett, Lixiao Zhu, Rafael Sousa Silva, Ryan Blake Jackson, and Tom Williams. Ipower: Incremental, probabilistic, open-world reference resolution. In Proceedings of the Annual Meeting of the Cognitive Science Society, 2022.

Experience

Employment

2021 Engineering/Code Intern, Attollo Prep, Lancaster, PA

Taught Python to a group of high school students from Lancaster County who belonged to underrepresented communities and backgrounds.


2018 - 2019 Student Reader, Franklin & Marshall College Computer Science Department, Lancaster, PA

Received, tested, and debugged anonymized code submissions in Python. Provided written feedback to the students in order to help them improve their programming skills.


External Referee Service

2022 Reviewer, Association for the Advancement of Artificial Intelligence (AAAI)


Undergraduate Research and Projects

2020 - 2021 AI Reasoning System, Franklin & Marshall Computer Science Department, Lancaster, PA

Assisted professor Justin Brody with his research project focused on the creation and simplification of an AI reasoning system based on reinforcement learning techniques.


2019 - 2020 Pulsar Matching, Franklin & Marshall Physics Department, Lancaster, PA

Developed an algorithm that filtered a dataset with over 40 thousand pulsar beams. The algorithm identifies which beams belonged to known pulsars and writes relevant information (e.g. declination and right ascension) to output files.


2018 Information Desk GUI, Individual Project, Lancaster, PA

Implemented a graphic user interface for information desk phone workers. The interface kept a log of college phone extensions and contained important information that was often requested by callers, such as directions to campus.

Honors and Awards

2021 John Kershner Scholar in Mathematics, Franklin & Marshall College Math Department, Lancaster

This honor is awarded to students of good and regular standing in the College for proficiency in mathematics, and includes an honorary stipend.


2020 Mathematics Honor Society Membership

Pi Mu Epsilon, a national mathematics honor society, with chapters located throughout the United States, elects members in post-secondary institutions according to their outstanding scholastic records and exceptional performance in mathematics. The society sponsors activities to engage members in mathematical and scholarly development.


2016 Opportunity Funds Program Scholarship

The Education USA Opportunity Funds program assists highly qualified students who are likely to be awarded full financial aid from U.S. colleges and universities but lack the financial resources to cover the up-front costs of obtaining admission, such as testing, application fees, or airfare. In addition to academic abilities, Opportunity Funds students bring cultural and socioeconomic diversity to U.S. campuses. I was one of the 23 students awarded from a pool of over 1000 applicants from all over Brazil.


2016 Gold Medalist, XIX Brazilian Astronomy and Astronautics Olympiad

2016 Bronze Medalist, I Physics Olympiad of Pernambuco

2016 Gold medalist, II Astronomy Olympiad of Pernambuco

2015 Gold medalist, I Astronomy Olympiad of Pernambuco

2013, 2014 Honorable Mention, Brazilian Mathematics Olympiad of Public Schools


Mentoring

Undergraduate Researchers

11/2022 - present Cailyn Smith - Impacts of latency on Performance of Autonomy.

06/2022 - present Michelle Lieng - Impacts of latency on Performance of Autonomy.

Technical Background

Proficient in: Python, Java, LaTeX, Microsoft Office, LibreOffice

Course knowledge in: C/C++, MySQL, Matlab

Familiar with: C# for Unity 2D, HTML, CSS, Javascript, R

References

  1. Dr. Tom Williams (Ph.D. Advisor)

Associate Professor of Computer Science, Colorado School of Mines, Golden, CO

E-mail: twilliams@mines.edu, office phone: (303) 273-3820