Literatures

Stuff I find to be an interesting read... discussing novel ideas in mathematics, neuroscience, machine learning, or their intersections.
My notes on literatures

Research Paradigm

• The Standard Model of Machine Learning
  Standardlizing all machine learning approahces, forming a standard model for ML.

• The Neuroconnectionist Research Programme
  Lakatosian research program setting computational understanding of the brain.

• Building Machines that Learn and Think like People
  Foundation review discussing potential research directions for human-like AI.

---

Reinforcement Learning

Theoretical Reinforcement Learning

• Maximum a Posteriori Policy Optimisation
  Probabilistic flavor-infused cutting-edge actor-critic.

• Proximal Policy Optimization Algorithms
  Actor-critic algorithm on steroids.

• IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
  Distributed RL framework for off-policy learning.

• Addiction as a Computational Process Gone Awry
  Using RL methods to model the addiction process.

---

Goal-Directed Deep Reinforcement Learning

• Divergent Representations of Ethological Visual Inputs Emerge from Supervised, Unsupervised, and Reinforcement Learning
  Proposes using ANNs to model the brain.

• Deep Neuroethology of a Virtual Rodent
  Aligning deep RL with biological counterparts.

• Whole-body Simulation of Realistic Fruit Fly Locomotion with Deep Reinforcement Learning
  Distributed trained MPO policy for goal-directed RL.

---

Inverse Kinematics Imitation Learning

• A Virtual Rodent Predicts the Structure of Neural Activity Across Behaviors
  Imitation learning mimicking rodent behaviors, showing similar neuronal activations.

• CoMic: Complementary Task Learning & Mimicry for Reusable Skills
  Encoder/decoder architecture transferring motor skills across tasks.

---

Representation Building

• Inductive Biases of Neural Network Modularity in Spatial Navigation
  Building beliefs in artificial agents through MOPDP conditions.

---

World Model & Agent Model

• Language Models Meet World Models: Embodied Experiences Enhance Language Models
  L-policy: building embodied world model into language model agent through embodied experiences and finetunning.

• Building Cooperative Embodied Agents Modularly with Large Language Models
  Building collaborative agent in an partially observable embodied environment.