Personal website
I’m a Postdoc in the Algorithms, Data Structures and Foundations of Machine Learning research group, headed by Prof. Kasper Green Larsen at Aarhus University (Department of Computer Science).
I did my Ph.D. at Inria centre at Université Côte d’Azur (concluded in 2023) under the supervision of Dr. Emanuele Natale.
I’m interested in understanding large, complex systems, typically through probabilistic methods. Currently, my research focuses on the theory of machine learning, particularly (statistical) learning theory and deep learning. I’m also interested in the empirical aspects of many fields, ranging from analogue computing to high-performance computing. In particular, I enjoy coding; at all levels: From algorithm design to low-level optimizations.
I hold a major in mathematics and a master’s degree in combinatorics from Universidade Federal do Ceará. There I worked with Prof. Fabricio Benevides on extremal and probabilistic combinatorics. I took part on a neuromorphic computing project while working for Hewlett Packard Enterprise.
da Cunha, A., Høgsgaard, M.M. and Larsen, K.G., 2024. Optimal Parallelization of Boosting. (Oral presentation) In Conference on Neural Information Processing Systems (NeurIPS). ArXiv
da Cunha, A., Larsen, K.G., and Ritzert, M., 2024. Boosting, Voting Classifiers and Randomized Sample Compression Schemes. Preprint. ArXiv
da Cunha, A., d’Amore, F. and Natale, E., 2023. Convolutional neural networks contain structured strong lottery tickets. In Conference on Neural Information Processing Systems (NeurIPS). OpenReview
da Cunha, A.C.W., d’Amore, F., Giroire, F, Lesfari, H., Natale, E., Viennot, L., 2023. Revisiting the Random Subset Sum problem. European Symposium on Algorithms (ESA). LIPIcs
da Cunha, A., Natale, E. and Viennot, L., 2022. Proving the Strong Lottery Ticket Hypothesis for Convolutional Neural Networks. In International Conference on Learning Representations (ICLR). OpenReview
Becchetti, L., da Cunha, A.C.W., Clementi, A., d’Amore, F., Lesfari, H., Natale, E. and Trevisan, L., 2022. On the Multidimensional Random Subset Sum Problem. Preprint. ArXiv
da Cunha, A.C.W., Natale, E. and Viennot, L., 2023. Neural Network Information Leakage Through Hidden Learning. In International Conference on Optimization and Learning (OLA). HAL
Da Cunha, A.C.W., Natale, E., Viennot, L., Institut national de recherche en sciences et technologies du numérique, 2022. Résistance équivalente modulable à partir de résistances imprécises (programmable equivalent resistances from imprecise resistors). France Patent deposit n°FR2210217.
Ambrosi, J.C., Da Cunha, A.C.W. and Cavalcante, J.R.A., Hewlett Packard Enterprise Development LP, 2021. System for a flexible conductance crossbar. U.S. Patent 11,200,948. Google Patents
Aguiar, G.D.S., Silveira, F.P., Lee, E.S., Antunes, R.J.D.R., Souza, J.G.D.C.E., Foltin, M., Cavalcante, J.R.A., Leite, L., Cunha, A.C.W.D., Frazao, M.V. and Trajano, A.F.R., Hewlett Packard Enterprise Development LP, 2022. SYSTEM AND METHOD FOR PROCESSING CONVOLUTIONS ON CROSSBAR-BASED NEURAL NETWORK ACCELERATORS FOR INCREASED INFERENCE THROUGHPUT. U.S. Patent Application 17/027,628. Google Patents