Celso M. de Melo

Computer Scientist
Associate Editor, IEEE Transactions on Affective Computing
Contact: celso.miguel.de.melo@gmail.com

Trends in Cognitive Sciences
1. Next-generation deep learning based on simulators and synthetic data
Deep learning models learn in a most artificial way: they require large quantities of labeled data to grasp even simple concepts. We highlight a solution to this challenge: synthetic data. Synthetic data are becoming accessible due to progress in rendering and generative pipelines. Paradoxically, this artificial solution will enable more natural learning, as seen in biological systems. Read the paper.
Natura Machine Intelligence
2. Incorporating physics into data-driven computer vision
This paper reviews methods to incorporate physics knowledge into deep learning pipelines. Read the paper.
Proceedings of IEEE
3. Social functions of machine emotional expressions
The research summarizes the social effects of emotion expressions in machines, including its role in building trust and promoting cooperation. Read the paper.
Frontiers in Robotics and AI
4. Risk of Injury Shapes Choices in Moral Dilemmas with AVs
Our shows that moral choices are shaped by risk of injury for involved parties in dilemmas involving autonomous machines. Read the paper.
Proceedings of the National Academy of Sciences U.S.A.
5. Human cooperation when acting through autonomous machines
Autonomous machines that act on our behalf are bound to face situations where individual interest conflicts with collective interest, and we must understand if people will cooperate when acting through them. We show, in the increasingly popular domain of autonomous vehicles, that people program machines to cooperate more than they would when acting directly with others. Read the paper.
Journal of Personality and Social Psychology
6. Reading People's Minds From Emotion Expressions
Emotion expressions can be windows into other people's minds. This research shows that people make inferences about how others are appraising the ongoing interaction from emotion expressions and, from this information, about others' beliefs, desires, and intentions. Read the paper.
Bootstrap Slider

My research focus is, broadly, in artificial intelligence, large pre-trained models, computer vision, and human-machine interaction. My research focuses on novel AI/ML paradigms enabled by the application of large pre-trained models to visual, robotic, and language tasks. Another portion of my research portfolio studies synthetic data for AI/ML, not only as a means to address the central challenge of acquiring high quality labeled data for training AI/ML, but as a means to enable a next generation of ML that can learn continually, multimodally, and interactively. A third focus of my research is on building socially intelligent machines, with a particular focus on affective computing.

I am a computer scientist with a research focus on artificial intelligence and human-machine interaction. I finished a postdoc at the USC Marshall School of Business. I earned my Ph.D. in Computer Science at the University of Southern California. This work was done at the Institute for Creative Technologies with Jonathan Gratch.I received a M.Sc. in Computer Science at the Technical University of Lisbon (IST) with Ana Paiva at the Synthetic Characters and Intelligent Agents Group (GAIPS).

Last updated: May 9, 2024