Symbolic Systems

Some resources for thinking about Universal Intelligence. Selective rather than exhaustive, biased to my taste, roughly ordered. Works dense with ideas highlighted. Resisted adding a Schmidhuber section.

---

Notes and Intuition

• How Complex Systems Fail

• Reflections on Trusting Trust

• Analogy as the Core of Cognition

• The Bitter Lesson

• Maxims of Signal Processing

• The Surprising Creativity of Digital Evolution: A Collection of Anecdotes from the Evolutionary Computation and Artificial Life Research Communities

• Why Philosophers Should Care About Computational Complexity

• On the Measure of Intelligence

• Is DNA Hardware or Software?

• Nothing makes sense in deep learning, except in the light of evolution

---

Models and Priors

• Attention Is All You Need

• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

• Universal Transformers

• Relational Inductive Biases, Deep Learning, and Graph Networks

• Scalable Training of Artificial Neural Networks with Adaptive Sparse Connectivity Inspired by Network Science

• Using Machine Learning to Replicate Chaotic Attractors and Calculate Lyapunov Exponents from Data

• Neural Turing Machines

---

Reinforcement Learning

• Programmatically Interpretable Reinforcement Learning
• Deep Reinforcement Learning Doesn’t Work Yet
• An Outsider’s Tour of Reinforcement Learning

---

Meta-Learning and Architecture Search

• HyperNetworks

• Evolving Reinforcement Learning Algorithms

• HOUDINI: Lifelong Learning as Program Synthesis

• RL²: Fast Reinforcement Learning via Slow Reinforcement Learning

• Learning to Learn by Gradient Descent by Gradient Descent

• Learning to Learn Without Gradient Descent by Gradient Descent

• Learning to Reinforcement Learn

• DARTS: Differentiable Architecture Search

• Meta Networks

• Lamarckian Evolution of Convolutional Neural Networks

---

Evolutionary Strategies

• Evolution Strategies as a Scalable Alternative to Reinforcement Learning

• How Learning Can Guide Evolution

• Cartesian Genetic Programming: Its Status and Future

• An Enhanced Hypercube-Based Encoding for Evolving the Placement, Density, and Connectivity of Neurons

• Neuroevolution of Self-Interpretable Agents

• State of Mutation Testing at Google

• On Connected Sublevel Sets in Deep Learning

---

Universal Constructors

• Von Neumann Universal Constructor

• An Implementation of von Neumann’s Self-Reproducing Machine

• Autoconstructive Evolution: Push, PushGP, and Pushpop

• SMCGP2: Self Modifying Cartesian Genetic Programming in Two Dimensions

• Self-Replicating and Self-Modifying Programs in Fraglets

• Backpropamine: Training Self-Modifying Neural Networks with Differentiable Neuromodulated Plasticity

• A Very Short Self-Interpreter

• A Neural Network That Embeds its Own Meta-Levels

  • An ‘Introspective’ Network That Can Learn To Run its Own Weight Change Algorithm
  • A ‘Self-Referential’ Weight Matrix
  • Steps Towards ‘Self-Referential’ Neural Learning: A Thought Experiment
  • Learning to Control Fast-Weight Memories: An Alternative to Recurrent Nets

---

Open-Ended Environments

• Open-Endedness: The Last Grand Challenge You’ve Never Heard of

• Wolfram Physics Project

• Minimal Criterion Coevolution: A New Approach to Open-Ended Search

• POET: Endlessly Generating Increasingly Complex and Diverse Learning Environments and their Solutions through the Paired Open-Ended Trailblazer

• Enhanced POET: Open-Ended Reinforcement Learning through Unbounded Invention of Learning Challenges and their Solutions

---

Multi-Agent Systems

• Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments

• Emergent Tool Use from Multi-Agent Interaction

• MAgent: A Many-Agent Reinforcement Learning Platform for Artificial Collective Intelligence

• Deal or No Deal? End-to-End Learning for Negotiation Dialogues

---

Artificial General Intelligence

• Gödel Machines: Self-Referential Universal Problem Solvers Making Provably Optimal Self-Improvements

• A Theory of Universal Artificial Intelligence based on Algorithmic Complexity

• Computable Variants of AIXI which are More Powerful than AIXItl

• Bad Universal Priors and Notions of Optimality

• AI-GAs: AI-Generating Algorithms, An Alternate Paradigm for Producing General Artificial Intelligence

---

Information Theory

• Randomness and Mathematical Proof

• Driven by Compression Progress: A Simple Principle Explains Essential Aspects of Subjective Beauty, Novelty, Surprise, Interestingness, Attention, Curiosity, Creativity, Art, Science, Music, Jokes

• Quantifying the Rise and Fall of Complexity in Closed Systems: The Coffee Automaton

• Normalized Information Distance

• The Information Theory of Individuality

• Computational Complexity as an Ultimate Constraint on Evolution

• The Algorithmic Origins of Life

• Automata and Graph Compression

• Top-Down Causation and Emergence: Some Comments on Mechanisms

• Self-Referencing Cellular Automata: A Model of the Evolution of Information Control in Biological Systems

• Information Closure Theory of Consciousness

---

Biology and Neuroscience

• What Can Information-Asymmetric Games Tell Us About the Context of Crick’s ‘Frozen Accident’?

• Neuronal Boost to Evolutionary Dynamics

• How Self-Organization Can Guide Evolution

• The Evolution of Cooperation by the Hankshaw effect.

• Breeding Novel Solutions in the Brain: a Model of Darwinian Neurodynamics

• How Evolution Learns to Generalise: Using the Principles of Learning Theory to Understand the Evolution of Developmental Organisation