Published in NeurIPS 2025, 2025
NeurIPS 2025 work on cost-aware LLM cascades that enforce probabilistic compute budgets while sustaining reasoned accuracy.
Download here
Published in ICML 2025, 2025
Introduces a Koopman-inspired linear recurrent model that matches transformer accuracy with drastically lower inference cost.
Download here
Published in arXiv preprint arXiv:2507.02076, 2025
Comprehensive 2025 survey of adaptive inference, conditional computation, and controllable test-time compute for large language models.
Download here
Published in AISTATS 2025, 2025
AISTATS 2025 framework for online deep learning that ensembles learners with adaptive exploration.
Download here
Published in Transactions on Machine Learning Research (TMLR), 2025
TMLR paper proposing incremental GNN recommenders with curated negative sampling that preserve personalization.
Download here
Published in NeurIPS Efficient Natural Language and Speech Processing Workshop (ENLSP-IV 2024), 2024
NeurIPS 2024 ENLSP-IV workshop paper on activating only the layers needed for decoder-only transformers at inference time.
Published in arXiv preprint arXiv:2410.11765, 2024
Introduces cluster-aware reweighting for graph neural networks tackling extreme class imbalance.
Download here
Published in IEEE Transactions on Visualization and Computer Graphics 30(8):5693-5704, 2023
IEEE TVCG work enabling high-quality human motion interpolation using transformer-based architectures.
Download here
Published in AAAI 2023, 2023
AAAI 2023 paper proposing structure-aware continual learning strategies for recommender models.
Download here
Published in IEEE Signal Processing Letters 31, 16-20, 2023
IEEE Signal Processing Letters article coupling normalizing flows with Population Monte Carlo for efficient Bayesian inference.
Download here
Published in 56th Asilomar Conference on Signals, Systems, and Computers, 2022
Asilomar 2022 paper that jointly infers GNN architecture, graph structure, and weights in a Bayesian setting.
Download here
Published in EUSIPCO 2022, 2022
EUSIPCO 2022 contribution aligning time-series signals across dynamic graphs via contrastive objectives.
Download here
Published in AAAI 2022, 2022
AAAI 2022 paper introducing Bag Graph, a Bayesian GNN formulation for multiple-instance learning.
Download here
Published in McGill University (Master's Thesis), 2021
Master’s thesis detailing graph-based representations for MIL problems with relational structure.
Download here
Published in Asilomar Conference on Signals, Systems, and Computers 2020, 2020
Asilomar 2020 paper extending GNNs to operate over nodes represented by distributions rather than single feature vectors.
Download here
Published in IETE Journal, 2019
This paper develops an SDR based satellite communication project for undergraduate EE students taking an Intro to Communication class.
Download here
Published in 2019 22nd International Conference on the Computation of Electromagnetic Fields (COMPUMAG), 2019
Rather than attempting to solve a common EM problem using Deep Learning, which has been done before, we focus on getting an extremely fast, but also accurate estimation.
Download here