Midterm 4 (2025)
Completion requirements
Assignment Rules and Execution
List of papers
Due: Friday, 27 June 2025, 6:00 PM
Assignment Rules and Execution
The fourth midterm covers the program until lecture 36. Differently from the previous ones, this midterm is based on reading and summarizing the main findings of a single paper chosen from the list provided below.
Students are expected to deliver a short presentation (no more than 8 slides) covering the following content:
- Introduction to the problem
- Model description
- Key catch of the model, represented by a commented equation
- Key (empirical) result
- Comment on novelties, strong points and weaknesses
I will pay particular attention to the technical depth and understanding of the paper which you will convey through point 3 above. As usual the presentation (in PDF please) should be upload here by the (strict) deadline.
List of papers
- Junyoung Chung, Sungjin Ahn, Yoshua Bengio, Hierarchical Multiscale Recurrent Neural Networks, ICLR 2017
- Bo Chang, Minmin Chen, Eldad Haber, Ed H. Chi, AntisymmetricRNN: A Dynamical System View on Recurrent Neural Networks, ICLR 2019
- Peters & Schaal, Reinforcement learning of motor skills with policy gradients, Neural Networks, 2008
- Schulman et al, Trust Region Policy Optimization, ICML, 2015
- Ho and Ermon, Generative Adversarial Imitation Learning, NIPS 2016
- Arjovsky, M., & Bottou, L. Towards principled methods for training generative adversarial networks. ICLR 2017
- Y. Song & S. Ermon, Generative Modeling by Estimating Gradients of the Data Distribution, NeurIPS 2019
- Jonathan Ho et al, Denoising Diffusion Probabilistic Models, NeurIPS 2020
- J. Austin, et al, Structured denoising diffusion models in discrete state-spaces, NeurIPS 2021
- Kingma & Dhariwal, P, Glow: Generative flow with invertible 1x1 convolutions, NeurIPS 2018
- G. Papamakarios et al, Masked Autoregressive Flow for Density Estimation, NeurIPS 2017
- Aditya Ramesh et al. “Hierarchical Text-Conditional Image Generation with CLIP Latents." arxiv Preprint arxiv:2204.06125 (2022)
- Andrea Ceni, Andrea Cossu, Maximilian W Stölzle, Jingyue Liu, Cosimo Della Santina, Davide Bacciu, Claudio Gallicchio, Random Oscillators Network for Time Series Processing, AISTATS 2024
- B. Scellier and Y. Bengio, “Equilibrium Propagation: Bridging the Gap between Energy-Based Models and Backpropagation,” Frontiers in Computational Neuroscience, vol. 11, 2017
- Petar Veličković, Rex Ying, Matilde Padovano, Raia Hadsell, Charles Blundell, Neural Execution of Graph Algorithms, ICLR 2020
- F. Errica, D. Castellana, D. Bacciu, A. Micheli, The Infinite CGMM, ICML 2022
- B. Chamberlain et al, Grand: Graph neural diffusion. ICML 2021
- A. Gravina, M. Eliasof, C. Gallicchio, D. Bacciu, & C.-B. Schönlieb On Oversquashing in Graph Neural Networks Through the Lens of Dynamical Systems. AAAI 2025, Full paper here: https://arxiv.org/pdf/2405.01009
- Phillip Lippe, Sara Magliacane, Sindy Löwe, Yuki M. Asano, Taco Cohen, Efstratios Gavves, CITRIS: Causal Identifiability from Temporal Intervened Sequences, ICML 2022
- R Massidda, F Landolfi, M Cinquini, D Bacciu, Constraint-Free Structure Learning with Smooth Acyclic Orientations, ICLR 2024
- Dao Tri, Albert Gu, Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality, ICLR 2024, https://arxiv.org/pdf/2405.21060