# Reading Group

We hold biweekly “reading-group” meetings (duration: 60 minutes). Every two weeks, a different group member takes the lead, presenting a chosen paper for discussion. Additionally, to the biweekly schedule, we frequently invite external speakers to give talks or tutorials.

## Upcoming

**December 02, 2024**@ 17:00 JST*Reading Group.* Emtiyaz Khan will moderate a paper discussion, topic TBA.

**November 18, 2024**@ 17:00 JST*Reading Group.* Christopher Anders will moderate a paper discussion, topic TBA.

**October 29, 2024**@ 11:00 JST*Seminar.* Oliver Eberle. “Analyzing the Inner Workings of Foundation Models: Towards Insights, Transparency, and AI Safety” doorkeeper

**October 07, 2024**@ 17:00 JST*Reading Group.* Yohan Jung will moderate a paper discussion. Paper 1, Paper 2.

## Past Meetings

**September 30, 2024***Reading group.* Mutual feedback on ICLR submissions.

**September 09, 2024***Reading group.* Kai Arulkumaran will give a talk with the title “Towards Brain-Controlled Robots”. doorkeeper.

**August 28, 2024***Discussion.* Internal group discussion (quarterly meeting). Note this reading-group is on a Wednesday!

**August 19, 2024***Reading Group.* JinYeong Bak. Memoria: Resolving Fateful Forgetting Problem through Human-Inspired Memory Architecture. doorkeeper

**June 06, 2024***Invited Talk.* François-Xavier Briol gives a talk on Robust and Conjugate Gaussian Process Regression. doorkeeper

**June 05, 2024***Invited Talk.* Patrick Shafto will give a talk on Mathematical foundations for learning agents. doorkeeper

**June 03, 2024***Invited Talk.* Tom Burns will give a talk on Semantically-correlated memories in a dense associative model. doorkeeper

**May 29, 2024***Invited Talk.* Talks by Fred Kunstner (Adaptive Methods in Machine Learning and Why Adam Works so Well) and Aaron Mishkin (Optimal Sets and Solution Paths of ReLU Networks). doorkeeper1, doorkeeper2

**May 27, 2024***Internal Meeting.* Anita introduces her research to the group.

## Show all reading group meetings

**May 20, 2024***Internal Discussion.* Mutual feedback on NeurIPS submissions.

**April 03, 2024***Talk.* Rafael Cabral: “A Bayesian workflow for efficient and automatic model checking and robustness analysis”. doorkeeper

**March 25, 2024***Reading group.* Zhedong will moderate a paper discussion on
GaLore: Memory-Efficient LLM Training by Gradient Low-Rank Projection.

**March 11, 2024***Reading group.*
Zwane Sicelukwanda
will give a talk on Safe Robot Motion Generation With Gaussian
Processes. doorkeeper

**February 26, 2024***Reading group.* Hugo will moderate a paper discussion on “A view of Estimation of Distribution Algorithms through the lens of Expectation-Maximization”.

**February 16, 2024***Reading group.*
Pratik Chaudhari gives a talk on ‘A Picture of the Prediction Space of Deep Networks’. doorkeeper

**January 29, 2024***Internal meeting.* Mutual feedback on the group’s ICML submissions.

**January 15, 2024***Internal meeting.* Catching up in the new year, discussion on
upcoming events and deadlines in 2024.

**December 04, 2023***Reading Group.* Keigo moderates a discussion on
Junk DNA-Hypothesis. The reading
group will be on a winter-break afterwards, and resume in the new year.

**November 20, 2023***Reading Group.* Thomas moderates a discussion on minimum description
length, deep-learning and Bayes.

**November 06, 2023***Reading Group.* Geoffrey moderates a discussion on A Geometric Interpretation of the Metropolis–Hastings Algorithm.

**October 24, 2023***Reading Group.* Dharmesh moderates a discussion on the paper On Masked Pre-training and the Marginal Likelihood.

**September 25, 2023***Discussion.* Mutual Feedback on ICLR submissions.

**September 11, 2023***Quarterly Internal Group Meeting.*

**September 01, 2023***Talk.* Geoffrey Wolfer: From Distributions to Markov Chains: Recent Advances in Inference and Geometry. [doorkeeper]

**August 28, 2023***Reading group.* Peter:
Studying Large Language Model Generalization with Influence Functions.

**August 21, 2023***Long Seminar.* Etash Guha gives a tutorial on conformal prediction. Naima Borras speaks about local learning.

**August 18, 2023***Talk.* So Takao: Improving data-assimilation for weather forecasting: A graph-based Bayesian perspective.

**July 03, 2023***Reading group.* Quarterly internal group meeting.

**June 12, 2023***Talk.* Graham Neubig Is My NLP Model Working? The Answer is Harder Than You Think. doorkeeper link

**May 22, 2023***Reading group.* Etash and Yuesong’s research overview.

**May 08, 2023***Reading group.* Joe will give a talk on memory.

**April 10, 2023***Reading group.* Negar will moderate a discussion on
Diffusion models as weighted ELBOs.

**March 27, 2023***Discussion.* Session on productive time management in academia.

**March 13, 2023***Reading group.* Gianma will moderate a discussion on implicit bias
of SGD.

**February 27, 2023***Reading group.* Thomas will talk about variational bounds.

**February 13, 2023**Internal group meeting.

**January 30, 2023**Informal meeting, chatting, catch-up.

**January 23, 2023**Mutual feedback on ICML papers.

**December 19, 2022***Reading Group*. Christmas reading group (paper to be decided, bring cookies).

**December 05, 2022***Reading Group.* Discussion about interesting papers from NeurIPS 2022 and ICLR 2023.

**November 21, 2022**Internal group meeting.

**November 14, 2022***Talk*. Keigo Nishida (Osaka University) will talk about his work on AdamB: Decoupled Bayes by Backprop With Gaussian Scale Mixture Prior.

**November 07, 2022***Reading Group*. Simplifying Model-based RL: Learning Representations, Latent-space Models, and Policies with One Objective.

**October 25, 2022***Reading Group*. Unusual day (Tuesday)! Discussion of the paper Machine Theory of Mind.

**October 17, 2022***Seminar*. Unusual time. Talk by Cuong N. Nguyen on Generalization Bounds for Deep Transfer Learning Using Majority Predictor Accuracy

**October 04, 2022***Seminar*. Unusual day and time (Tuesday). Talk by Karim Lounici on Meta-Learning Representations with Contextual Linear Bandits

**September 26, 2022***Reading Group*. Discussion of the paper Formalizing the Generalization-Forgetting Trade-off in Continual Learning.

**September 12, 2022***Reading Group*. Discussion of the paper Fortuitous Forgetting in Connectionist Networks.

**August 29, 2022***Reading Group*. Discussion of recent papers on natural-gradient descent Paper 1, Paper 2, Paper 3.

**August 15, 2022***Seminar*. Talk by David Tomàs Cuesta on his lab rotation project.

**July 15, 2022***Reading Group*. Unusual day and time (Friday)!! Discussion of notable ICML, ICLR and COLT papers.

**July 14, 2022***Talk*. Unusual day and time (Thursday). Presentation by Itay Evron on “How catastrophic can catastrophic forgetting be in linear regression?”

**June 20, 2022**Internal group meeting.

**May 23, 2022***Seminar*. Talk by Geoffrey Wolfer on Inference in Markov Chain from a Single Finite Trajectory.

**May 09, 2022***Seminar*. Tojo’s term project presentation and research discussion on continual RL.

**April 25, 2022***Reading Group*. Discussion of the Paper by Hassabis et al., Neuroscience-Inspired Artificial Intelligence.

**April 11, 2022***Seminar*. Talk by Vincent Fortuin. Title: On the Importance of Priors in Bayesian Deep Learning.

**March 28, 2022***Reading group*. Discussion of AISTATS 2022 papers.

**March 14, 2022***Reading Group*. Discussion of the Paper by Bubeck and Sellke,
Universal Law of Robustness via Isoperimetry.

**February 28, 2022***Reading Group*. Discussion of the paper
Deep Learning through the Lens of Example Difficulty.

**February 21, 2022***Reading Group*. Discussion of the paper by Kaplan and Friston,
Planning and Navigation as Active Inference.

**February 17, 2022***Seminar*. Talk by Lionel Riou-Durand. Title: Metropolis Adjusted Underdamped Langevin Trajectories: a robust alternative to Hamiltonian Monte-Carlo.

**February 14, 2022***Reading Group*. Discussion of the paper
A Survey of Exploration Methods in Reinforcement Learning.

**February 07, 2022**Internal group meeting.

**January 31, 2022***Reading Group*. Discussion of the paper [Connections and Equivalences between the Nyström Method and Sparse Variational Gaussian Processes].

**January 24, 2022***Reading Group*. Discussion of the paper [Information-Theoretic Generalization Bounds for Stochastic Gradient Descent].

**January 17, 2022**After holiday catch-up and discussion on well-being in ML.

**December 20, 2021***Reading Group*. Discussion of the paper [The Seven Tools of Causal Inference with Reflections on Machine Learning].

**December 06, 2021***Seminar*. Term project talk by Ted Tinker.

**November 29, 2021***Seminar*. Talk by Sébastien Loustau. Title: Deep learning theory for power-efficient algorithms. Slides.

**November 22, 2021***Reading Group*. Discussion of the paper [Active Bias: Training More Accurate Neural Networks by Emphasizing High Variance Samples].

**November 15, 2021***Reading Group*. Discussion of the paper [Dual Parametrization of Sparse Variational Gaussian Processes].

**November 09, 2021***Seminar*. Talk by Ilsang Ohn on Adaptive variational Bayes: Optimality, computation and applications. Register via Doorkeeper.

**November 01, 2021***Reading Group*. Discussion of the paper [The Geometry of Abstract Learned Knowledge in the Hippocampus].

**October 18, 2021***Reading Group*. Discussion of the paper by van der Hoeven et al.: [The Many Faces of Exponential Weights in Online Learning].

**October 11, 2021**Internal group meeting.

**October 04, 2021***Reading Group*. Discussion of the paper by Khan and Swaroop, 2021: [Knowledge-Adaptation Priors]

**September 27, 2021***Reading Group*. Discussion of the paper by Pearce et al., 2021: [Understanding Softmax Confidence and Uncertainty]

**September 13, 2021***Reading Group*. Discussion of the paper by Hara et. al., 2019: [Data Cleansing for Models Trained with SGD]

**September 06, 2021***Reading Group*. Discussion of the paper by Khan and Rue, 2021: [The Bayesian Learning Rule]

**August 30, 2021***AIP Seminar*. Talk by Jiaxin Shi: “Sampling with
Mirrored Stein Operators”. [Paper]
[Doorkeeper]

**August 30, 2021**Casual chat in gather.town.

**August 23, 2021***Reading Group*. Discussion of the paper by Li et al., 2019: [Hessian based analysis of SGD for Deep Nets: Dynamics and Generalization]

**August 16, 2021***Reading Group*. Discussion of the paper by Tsividis et al., 2021: [Human-Level Reinforcement Learning through Theory-Based Modeling, Exploration and Planning]

**August 09, 2021***Reading Group*. Discussion of the paper by Hobbhahn and Hennig (2021): [Laplace Matching for Fast Approximate Inference in Generalized Linear Models]

**August 02, 2021***Seminar*. Leaving presentation by Dharmesh Tailor

**July 26, 2021***Reading group*. Discussion of the paper by Raj, Musco and Mackey (2020): [Importance Sampling via Local Sensitivity]

**July 19, 2021***Reading group*. Discussion of ICML 2021 papers.

**July 12, 2021***Reading group*. Discussion of the paper by Azoury and Warmuth (2001): [Relative Loss Bounds for On-Line Density Estimation with the Exponential Family of Distributions]

**July 05, 2021***Reading group*. Discussion of several knowledge distillation papers: [Paper 1], [Paper 2], [Paper 3]

**June 28, 2021**Social chat in gathertown.

**June 22, 2021***AIP seminar*. Talk by Jooyeon Kim: “Stochastic optimal control and probabilistic modeling for detecting and reducing fake news”

**June 07, 2021**Internal group meeting.

**May 31, 2021***Seminar*. Talk by Julyan Arbel. Title: Approximate Bayesian computation with surrogate posteriors [Paper]

**May 17, 2021**Mutual feedback on NeurIPS papers.

**May 10, 2021***Reading group*. Discussion of ICLR 2021 papers.

**April 26, 2021***Reading group*. On the origin of implicit regularization in stochastic gradient descent. [Paper]

**April 19, 2021***Reading group*. Discussion of AISTATS 2021 papers.

**April 12, 2021***Talk*. Talk by Dhruva Tirumala: Behavior Priors for Efficient Reinforcement Learning. [Paper]

**March 29, 2021***Talk + Tutorial*. Talk by Fariz Ikhwantri on “Knowledge distillation with DNN2GP” + tutorial by Thomas Möllenhoff on convex duality.

**March 22, 2021**Informal group discussion.

**March 16, 2021***Seminar, unusual day and time!* Talk by Daniele Calandriello: “Scalable Determinantal Point Processes for Machine Learning”.

**March 01, 2021**Internal group meeting.

**February 22, 2021***Seminar*. Talk by Andrew Foong and David Burt: “On the Expressiveness of Approximate Inference in Bayesian Neural Networks”. [Paper]

**February 15, 2021**Talk by Khimya Khetarpal and Matt Riemer: “Towards Continual Reinforcement Learning: A Review and Perspectives”. [Paper]

**February 08, 2021**Well-being chat.

**February 01, 2021***Seminar*. Talk by Blair Bilodeau and Jeffrey Negrea: “Relaxing the I.I.D. Assumption: Adaptively Minimax Optimal Regret via Root-Entropic Regularization”. [Paper]

**January 25, 2021**Mutual feedback on ICML submissions.

**January 18, 2021**Catch up meeting.

**December 17, 2020***Seminar, unusual day!* Erik.

**November 30, 2020***Reading group*. Qi Qian, Hao Li, Juhua Hu: “Efficient Kernel Transfer in Knowledge Distillation”. [Paper]

**November 16, 2020***Seminar*. Peter, Thomas. Uncertainty estimation for Bayesian Neural Networks using infinite-width nets.

**November 06, 2020**Pre-ICLR paper discussion.

**October 30, 2020***Reading group*. Ben Recht: “A Tour of Reinforcement Learning: The View from Continuous Control”. [Paper]

**October 23, 2020**Team work review for the past 4 months. Siddarth’s talk on functional regularization on the memorable past. [Paper]

**October 22, 2020**Unusual day and time! Dimitri Meunier: “Meta Learning meets Variational Inference. Learning priors with guarantees”.

**October 09, 2020**Feedback discussion on the new team webpage.

**September 25, 2020**Talk by François-Xavier Briol: “Stein’s Method for Computational Statistics and Machine Learning”. [Slides]

**September 18, 2020**Research chat and virtual breakfast/lunch/dinner in gather.town.

**August 28, 2020**Lab members project overview and discussion.

**August 21, 2020**Pierre’s grant presentation rehearsal and feedback.

**August 07, 2020***Reading group.* “Generalized Variational Inference: Three arguments for deriving new posteriors”. [Paper]

**July 17, 2020**Whole week: comments on ICML tutorials and talks.

**July 10, 2020**Pre-ICML paper discussion.

**June 26, 2020**Talk by Giulia Denevi on Efficient Lifelong Learning Algorithms: Regret Bounds and Statistical Guarantees.

**June 12, 2020**Internal kickoff meeting.

**June 05, 2020**Talk by Peter Nickl on “Variational Bayes for Infinite Mixtures of Local Regressors with Robotics Applications”. [Thesis]

**May 29, 2020**Talk by Gian Maria Marconi on Manifold Regression by Structured Prediction: methodology and applications. [Paper]

**May 22, 2020**Feedback discussion on our NeurIPS submissions.

**May 15, 2020**Talk by Thomas Möllenhoff on Flat Metric Minimization with Applications in Generative Modeling. [Paper]

**May 01, 2020***Reading Group*. “Continual Deep Learning by Functional Regularisation of the Memorable Past”. [Paper]

**April 24, 2020***Reading Group*. Shalev-Shwartz: “Introduction to online learning”, Chapter 2 (end). [Paper]