Readings

Motivation

Meta-learning according to Schimidhuber Blog

*On the Measure of Intelligence Arxiv2019

Deep-learning

The Little Book of Deep Learning (Univ. De Geneve, 2023)

An Overview of Deep Learning for Curious People Lil’Log Blog 2017

The Transformer Family Version 2.0 Lil’Log Blog 2023

Basics of Meta-learning: OptimisationMetric/Model -based

*Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks ICML2017

*Conditional Neural Processes ICML2018

*Matching Networks for One Shot Learning NeurIPS2016

Advanced Meta-learning

*TabPFN: A Transformer That Solves Small Tabular Classification Problems in a Second ICLR2023

*MotherNet: A Foundational Hypernetwork for Tabular Classification Arxiv2023

*Dynamic Weighted Majority: An Ensemble Method for Drifting Concepts JMLR2007

*Supermasks in Superposition NeurIPS2020

*What's Hidden in a Randomly Weighted Neural Network? CVPR2020

*Deep Incremental Learning for Financial Temporal Tabular Datasets with Distribution Shifts Arxiv2023

*Meta-learning from Tasks with Heterogeneous Attribute Spaces NeurIPS2020

Meta-learning using Generative Language Models 

*Meta-learning via Language Model In-context Tuning ACL2022

*Transformers learn in-context by gradient descent ICLR2023

*Do Pre-trained Transformers learn In-Context by Gradient Descent? ICML2024

*Learning to Learn via Self-Critique NeurIPS2019

*Meta-Learning via Classifier(-free) Diffusion Guidance Arxiv2023

Neuro-symbolic Program Synthesis

*DreamCoder: Wake-sleep Bayesian program learning NeurIPS2020

*Neural networks for abstraction and reasoning: Towards broad generalization in machines Arxiv2024

*Addressing the Abstraction and Reasoning Corpus via Procedural Example Generation Arxiv2024

*CodeIt: Self-Improving Language Models with Prioritized Hindsight Replay Arxiv2024

*Combining Transduction and Induction for Abstract Reasoning

Hypothesis Search: Inductive Reasoning with Language Models Arxiv2024

Graphs, Constraints, and Search for the Abstraction and Reasoning Corpus

Generalized Planning for the Abstraction and Reasoning Corpus Arxiv2024

Bayesian Program Learning by Decompiling Amortized Knowledge PMLR2024

Note: while covering each paper we may need to also cover some key references therein; these are not listed above. Further the list of papers are likely to change continuously. Those papers that are required reading for exams will be marked by a * well in time.