Luciano Dyballa
About Luciano Dyballa
Luciano Dyballa is a Postdoctoral Associate at Yale University, specializing in computer science and artificial intelligence. He holds a PhD from Yale and has contributed to significant advancements in machine learning through his research and mentorship of undergraduate students.
Work at Yale University
Currently, Luciano Dyballa serves as a Postdoctoral Associate at Yale University, a position he has held since 2021. His previous roles at Yale include Graduate Student Researcher from 2015 to 2021 and Graduate Teaching Assistant from 2016 to 2020. During his tenure, he has contributed to various research initiatives and has been involved in mentoring undergraduate students on their senior-year projects. Dyballa has also established interdisciplinary collaborations across six universities, organizing weekly meetings to share research findings.
Education and Expertise
Luciano Dyballa's educational background includes a Doctor of Philosophy (PhD) in Computer Science from Yale University, which he completed from 2015 to 2021. Prior to this, he earned a Master of Science (MS) in Computer Science from the Federal University of Rio de Janeiro from 2013 to 2015. His undergraduate studies include a Bachelor of Science (BS) in Chemical Engineering from the same university. Dyballa's expertise lies in artificial intelligence and machine learning, with a focus on image classification and dimensionality reduction.
Professional Experience at Petrobras
Before his academic career, Luciano Dyballa worked as a Process Engineer at Petrobras in Rio de Janeiro, Brazil, from 2010 to 2013. In this role, he applied his engineering knowledge to various processes within the company, gaining practical experience in the field. This early professional experience contributed to his analytical skills and understanding of complex systems.
Research Contributions and Publications
Luciano Dyballa has made significant contributions to the field of artificial intelligence and machine learning. He engineered a zero-shot learning approach for image classification, demonstrating that intermediate layers can generalize up to 50% better than the output layer. Additionally, he developed a Python-based machine learning package for dimensionality reduction, achieving a 15% increase in accuracy. Dyballa has published research papers in high-impact conferences and journals, including two at ICLR and two at PNAS, focusing on advancements in AI and machine learning.
Community Engagement and Mentorship
Throughout his academic career, Luciano Dyballa has actively engaged in community and mentorship activities. He has mentored four undergraduate students on their senior-year projects, providing guidance and support in their research endeavors. His commitment to fostering collaboration is evident in his establishment of interdisciplinary partnerships across six universities, where he organized weekly meetings to facilitate the exchange of research findings.