I am a 2nd year PhD student on the Dual Degree program between Carnegie Mellon University & Instituto Superior Técnico, supervised by Prof. Lei Li and Prof. Arlindo Oliveira.
Before starting my PhD, I completed my undergraduate studies in Electrical and Computer Engineering and a master's in Data Science, both at Instituto Superior Técnico.
My research in mainly focused on Security and Privacy of GenAI models, with an enphasis on Membership Inference Attacks. In other words: Trying to figure out wheter specific data was used to train a certain model!
I'm always happy to connect and collaborate. If you work in a similar area or see potential for us to work together, feel free to reach out :)
Action required
Problem: The current root path of this site is "baseurl ("_config.yml.
Solution: Please set the
baseurl in _config.yml to "News
Priberam Machine Learning Seminar - University of Lisbon: Instituto Superior Técnico
Poster Session: Information Retrieval and TextMining 3, 16:00-17:30, November 14.
Poster Session: Hall C 4-9, 13:30-15:00, 25 July.
Education
-
Carnegie Mellon University and Instituto Superior TécnicoDual Degree Ph.D. in Language Technologies and Computer ScienceFeb. 2024 - presentCurrently in Lisbon, Portugal
-
Instituto Superior TécnicoMSc. in Data Science and EngineeringSet. 2020 - Dec. 2022Lisbon, Portugal
-
Instituto Superior TécnicoBSc. in Electrical and Computer EngineeringSet. 2017 - Dec. 2020Lisbon, Portugal
Selected Publications (view all )
DIS-CO: Discovering Copyrighted Content in VLMs Training Data
André V. Duarte, Xuandong Zhao, Arlindo L. Oliveira, Lei Li
Under review, 2025
DIS-CO identifies copyrighted content in VLMs training data by showing that models can link movie frames to their titles in a free-form text generation setting, even when the frames are highly challenging, suggesting prior exposure during training.
DIS-CO: Discovering Copyrighted Content in VLMs Training Data
André V. Duarte, Xuandong Zhao, Arlindo L. Oliveira, Lei Li
Under review, 2025
DIS-CO identifies copyrighted content in VLMs training data by showing that models can link movie frames to their titles in a free-form text generation setting, even when the frames are highly challenging, suggesting prior exposure during training.
LumberChunker: Long-Form Narrative Document Segmentation
André V. Duarte, João Marques, Miguel Graça, Miguel Freire, Lei Li, Arlindo L. Oliveira
Findings EMNLP 2024
LumberChunker is a document segmentation method using LLMs to enhance retrieval by creating contextually coherent, variable-sized content chunks.
LumberChunker: Long-Form Narrative Document Segmentation
André V. Duarte, João Marques, Miguel Graça, Miguel Freire, Lei Li, Arlindo L. Oliveira
Findings EMNLP 2024
LumberChunker is a document segmentation method using LLMs to enhance retrieval by creating contextually coherent, variable-sized content chunks.
DE-COP: Detecting Copyrighted Content in Language Models Training Data
André V. Duarte, João Marques, Miguel Graça, Miguel Freire, Lei Li, Arlindo L. Oliveira
Proceedings of ICML 2024 ; Best Scientific Paper Award at Responsible AI Forum
DE-COP is a novel method to identify copyrighted content present in LLM training datasets. It works by showing that a model can recognize exact text excerpts if they were seen during training. It is applicable to models with/without logit outputs.
DE-COP: Detecting Copyrighted Content in Language Models Training Data
André V. Duarte, João Marques, Miguel Graça, Miguel Freire, Lei Li, Arlindo L. Oliveira
Proceedings of ICML 2024 ; Best Scientific Paper Award at Responsible AI Forum
DE-COP is a novel method to identify copyrighted content present in LLM training datasets. It works by showing that a model can recognize exact text excerpts if they were seen during training. It is applicable to models with/without logit outputs.