Sponsor: SAP Labs
Role: Principal Investigator
Description: This project applies Automated Planning and other AI techniques to help improve SAP's DataHub tools for the next generation of citizen data scientists.
Sponsor: Google Latin America
Role: Principal Investigator
Description: While deep learning techniques provide ever improving tools for classification tasks in most medical areas for which data is available, high classification performance provides virtually no insights to medical researchers about the underlying conditions. This project addressed this problem by advancing the state of the art on network visualization techniques for convolutional neural networks and providing meaningful and expert-backed insights into the condition being classified. These visualization techniques provide a tool for neuroscientists and other researchers to pry deeper into the neurological conditions they are studying.
Publications: TBA
Sponsor: SAP Labs
Role: Co-Principal Investigator
Description: This project investigated techniques to index very large databases, including techniques for unstructured databases. My key contribution to this project included AI techniques for self-driving databases and automatic indexing decisions based in reinforcement learning.
Publications: TBA
Sponsor: FAPERGS - State Government
Role: Co-Investigator
Description: this project developed goal recognition algorithms that deal with real world data (i.e. image and unstructured data). As part of the project created an annotated dataset of objects and relations between them suitable for goal recognition.
Publications: TBA
Sponsor: Google Latin America
Role: Principal Investigator
Description: in this project we developed approaches to extract norms from contracts in natural language and reason about them by identifying deontic conflicts, providing the first steps towards contract automation.
Publications: TBA
Sponsor: HP Inc.
Role: Principal Investigator
Description: in this project we developed goal recognition approaches for unstructured data aimed at inferring user intent in controlled environments to facilitate teamwork.
Publications: TBA
Sponsor: CNPq - Federal Government
Role: Co-Investigator
Description: this project developed integrated approaches for autonomous vehicles and sensor networks for remote data acquisition for the prevention and mitigation of natural disasters.
Publications: TBA
Sponsor: Dell Computer
Role: Principal Investigator
Description: in this project we developed machine learning algorithms to predict dispatch orders narrowed down by the location of the dispact.
Publications: TBA
Sponsor: Motorola / Lenovo
Role: Co-Investigator
Description: this project developed a number of deep learning techniques aimed at handling data from and about mobile phones. Our contribution included sentiment algorithms for analysis for mobile phone reviews, and identification of the sentiment about specific aspects of a mobile phone.
Publications: TBA
Sponsor: CNPq - Federal Government
Role: Principal Investigator
Description: this project was a partnership between PUCRS and Carnegie Mellon to develop and deploy robotic boats from Playtpus to monitor water quality in the city of Porto Alegre.
Publications: TBA
Sponsor: CAPES OBEDUC - Federal Government
Role: Principal Investigator
Description: the ACERTA project is an umbrella project about the diagnosis and prognosis estimation of dyslexia and dyscalculia done by a multidisciplinary team of researchers. My team in the School of Informatics works on applying stochastic graphical models in the diagnosis and prognosis of these learning disorders.
Publications: TBA
Sponsor: State Government
Role: Co-Principal Investigator
Description: this project was a partnership between the state government of Rio Grande do Sul and a local robotics company to develop robots and semi-autonomous algorithms to perform maintenance in oil storage facilities.
Publications: TBA
Sponsor: CNPq - Federal Government
Role: Principal Investigator
Description: this project developed a stochastic formal model of normative agent reasoning compatible with BDI agent programming language, as well as an enforcement mechanism, in order to carry out a simulation-based evaluation of various norm-enforcement policies against multiple agent behaviour types.
Publications: TBA
Sponsor: Samsung Brazil and the Brazilian Ministry of Science and Technology
Role: Co-Investigator
Description: This project investigated multiagent techniques to facilitate teams of humans to interact and collaborate across organizations. We developed a number of techniques for software development based on agent programming languages, multiagent interaction, cooperation and delegation protocols, as well as novel goal recognition techniques.
Publications: TBA
Sponsor: FAPERGS - State Government
Role: Co-Investigator
Description: this project developed reasoning mechanisms in agents operating in normative environments and develop algorithms and techniques to enable behaviour modification at runtime in response to dynamically changing norms.
Publications: TBA