Tomaso Poggio
Professor
Studies the theory of learning under physical, computational, and biological constraints, to understand the brain and build better machines.
Tomaso Poggio's lab at MIT works on computational neuroscience & cognition and learning theory & optimization. 28 papers since 2023, 8 at top venues, mostly ICML and NeurIPS. 137 alumni. Applications close .
At a glance
- papers, 2023–2026
- 7/ year
- at T1 venues
- 29%
- PhD students
- 0
- MS students
- 0
- alumni listed
- 137
Papers by yearfrom the lab homepage, 2023–2026
- 9 papers2026 · T1 3 · collecting
- 6 papers2025 · T1 2
- 3 papers2024 · T1 1
- 10 papers2023 · T1 2
T1 venuesOtherVenues: ICML, NeurIPS, ICLR
Research focuspaper titles that use each topic's words
- Learning theory & optimization6 papers
- Computational neuroscience & cognition0 papers
- brain0 papers
- cognitive sciences0 papers
- constraints0 papers
Paper list
From the lab homepage.
20269 papers
- CBMM PublicationA Perspective: Sparse Compositionality and Efficiently Computable Intelligence (opens in a new tab)
- ICLR (T1)A Universal Compression Theory: Lottery Ticket Hypothesis and Superpolynomial Scaling Laws (opens in a new tab)
- AI for Science Workshop at ICMLAgent Systems for Academic Research Automation (opens in a new tab)
- CBMM PublicationAssociative Memory as the Core of Intelligence in Technology and Evolution (opens in a new tab)
- High-dimensional Learning Dynamics (HiLD) Workshop at ICMLEdge of Stability Selectively Shapes Learning Across the Data Distribution (opens in a new tab)
- ICML (T1)Momentum Further Constrains Sharpness at the Edge of Stochastic Stability (opens in a new tab)
- High-dimensional Learning Dynamics (HiLD) Workshop at ICMLStabilizing Continuous-Time Kolen–Pollack Learning with a Scale-Balance Condition (opens in a new tab)
- High-dimensional Learning Dynamics (HiLD) Workshop at ICMLToo Sharp, Too Sure: When Calibration Follows Curvature (opens in a new tab)
- ICML (T1)Ubiquity of Hebbian Dynamics in Complex Learning Rules (opens in a new tab)
20256 papers
- ICLR (T1)Formation of Representations in Neural Networks (opens in a new tab)
- CBMM PublicationMultiplicative Regularization Generalizes Better Than Additive Regularization (opens in a new tab)
- Proceedings of the International Conference on Algorithmic Learning Theory (PMLR 272), 921-936On Generalization Bounds for Neural Networks with Low Rank Layers (opens in a new tab)
- Proceedings of the International Conference on Machine Learning (PMLR 267), 81199-81210Position: A Theory of Deep Learning Must Include Compositional Sparsity (opens in a new tab)
- Conference on Parsimony and LearningSGD with Weight Decay Secretly Minimizes the Ranks of Your Neural Networks (opens in a new tab)
- NeurIPS (T1)Training the Untrainable: Introducing Inductive Bias via Representational Alignment (opens in a new tab)
20243 papers
202310 papers
- CBMM PublicationA Homogeneous Transformer Architecture (opens in a new tab)
- 272Cervelli menti algoritmi (opens in a new tab)
- Research . doi:10.34133/research.0024Dynamics in Deep Classifiers trained with the Square Loss: normalization, low rank, neural collapse and generalization bounds (opens in a new tab)
- CBMM PublicationFeature learning in deep classifiers through Intermediate Neural Collapse (opens in a new tab)
- Analysis and Applications 21, 193For interpolating kernel machines, minimizing the norm of the ERM solution maximizes stability (opens in a new tab)
- CBMM PublicationNorm-Based Generalization Bounds for Compositionally Sparse Neural Networks (opens in a new tab)
- NeurIPS (T1)Norm-based Generalization Bounds for Sparse Neural Networks (opens in a new tab)
- CBMM PublicationSGD and Weight Decay Provably Induce a Low-Rank Bias in Deep Neural Networks (opens in a new tab)
- ICML (T1)System Identification of Neural Systems: If We Got It Right, Would We Know? (opens in a new tab)
- CBMM PublicationThe Janus effects of SGD vs GD: high noise and low rank (opens in a new tab)
Where alumni went137 listed
Apply to MITFall 2027
MIT EECS Computer Science PhD
- Round
- Single round
- Opens
- —
- GRE
- See program page
- English
- TOEFL/IELTS
- Fee
- USD 90
Confirmed on the program page.
Similar labsnearest by research summary
- CNSJTU
Dequan Wang
13 T1 - USUCSD
Yu-Xiang Wang
- Reinforcement learning
- USNYU
Qi Lei
- Learning theory & optimization
- CNZhejiang
Haobo Wang
78 T1 - CHETH Zurich
Niao He
- Learning theory & optimization
- CHEPFL
Pierre Vandergheynst
- Signal processing & information theory
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
MITcloses