Miguel Vasco is a PhD student in Computer Science at Tech Lisbon since 2018 and a researcher of the Group on Artificial Intelligence for People and Society (GAIPS) since 2016. Miguel aims at merging the topics of Artificial Intelligence and Human Cognition with Deep Learning. He will certainly be successful in such task.
Humans (and some robots) are able to perceive their environment through multiple sensory channels and obtain multimodal data from the world. From that data, they create rich internal representations of their environment in order to be able to act upon the world. Miguel likes to build biological-inspired generative models to learn multimodal representations. 45% of the time, they work every time.
Besides working hard, Miguel likes to make people laugh, pretend he is a Shaolin warrior, compose music and write descriptions of himself in third-person.
PhD Student in Computer Science, 2018-2022
Tech Lisbon, University of Lisbon
MSc in Engineering Physics, 2016
Tech Lisbon, University of Lisbon
In this work we explore the use of latent representations obtained from multiple input sensory modalities (such as images or sounds) in allowing an agent to learn and exploit policies over different subsets of input modalities. We propose a three-stage architecture that allows a reinforcement learning agent trained over a given sensory modality, to execute its task on a different sensory modality-for example, learning a visual policy over image inputs, and then execute such policy when only sound inputs are available. We show that the generalized policies achieve better out-of-the-box performance when compared to different baselines. Moreover, we show this holds in different OpenAI gym and video game environments, even when using different multimodal generative models and reinforcement learning algorithms.
Humans interact in rich and diverse ways with the environment. However, the representation of such behavior by artificial agents is often limited. In this work we present textitmotion concepts, a novel multimodal representation of human actions in a household environment. A motion concept encompasses a probabilistic description of the kinematics of the action along with its contextual background, namely the location and the objects held during the performance. Furthermore, we present Online Motion Concept Learning (OMCL), a new algorithm which learns novel motion concepts from action demonstrations and recognizes previously learned motion concepts. The algorithm is evaluated on a virtual-reality household environment with the presence of a human avatar. OMCL outperforms standard motion recognition algorithms on an one-shot recognition task, attesting to its potential for sample-efficient recognition of human actions.