Wilka Carvalho
Aspiring Cognitive Scientist

Knowledge of the Brain Informing Machine Learning and Vice-Versa

I have just joined Yan Liu’s research group and gave a small presentation this monday surveying current work using knoweldge about the brain to inform machine learning and vice-versa.

The focus of the presentation was a recent paper published by Google DeepMind in Nature, Human-level control through deep reinforcement learning.

The presentation was meant to introduce the group to my research interests. I chose to focus on research that has explored the relations between the architecture and functionality of the brain and artificial neural network models.

As I prepared for the presentation, I came across many interesting papers, articles, and videos. I thought I’d gather and share them here for others with similar interests.

Note: some things weren’t particularly relevant to my research interests but I found them really cool so I added them to this list anyway.

Papers

Human-level control through deep reinforcement learning
  1. Playing Atari with Deep Reinforcement Learning . This was the original model they created.
  2. Review by Jürgen Schmidhuber. Two members of the original DeepMind team worked in his lab. He claims some of the results of the paper had already been found by his lab.
  3. Code for Human-Level Control
  4. Play it again: reactivation of waking experience and memory. The memory consolidation described here was the inspiration for their “action replay” model.
Deep architecture possibly used by brain for vision
  1. A quantitative theory of immediate visual recognition. This is a quantitative model for how the brain performs rapid object recognition. It indicated that the brain had a deep architecture.
  2. Sparse belief net model for V2. This was an ANN model which successfully replicated results from visual cortices V1 & V2.
  3. Learning Deep Architectures for AI, Shallow vs. Deep Sum-Product Networks. Two articles by Yoshua Bengio in which he explores deep architectures. He claims deep architectures may be necessary to learn the complicated functions necessary to represent high-level abstractions, e.g. vision, language, etc.
“Neural-Turing Machines”
  1. The Human Turing Machine: a Neural Framework for Mental Programs
  2. Neural Turing Machines
Misc. Papers
  1. Modha, Dharmendra S et al. “Cognitive Computing.” Communications of the ACM 54.8 (2011): 62–71. Couldn’t get a link. This describes IBM’s brain-inspired chip and some of the ways software is being implemented for it.
  2. Is the Brain a Good Model for Machine Intelligence?. Fun series of articles discussing advancements and limitations in modeling the brain’s computations.
  3. Machines That Think for Themselves
  4. Learning to Execute. A neural network to learn simple computer programs.
  5. RM-SORN: a reward-modulated self-organizing recurrent neural network. A dynamic neural network model that uses Hebbian learning to learn to perform a variety of tasks.
  6. Show and Tell: A Neural Image Caption Generator. An article about transfer learning (learning being transferred from one network to another)
  7. Unsupervised and Transfer Learning Challenge: a Deep Learning Approach
  8. Unsupervised feature learning for audio classification using convolutional deep belief networks. Deep learning applied to audio.
  9. A Large-Scale Model of the Functioning Brain. A mini-brain model that can redraw images it is presented with - seems pretty cool.
  10. Survery of papers on transfer learning
  11. Deep Learning Reading List

Other

Articles
  1. Deep Minds: An Interview with Google’s Alex Graves & Koray Kavukcuoglu
  2. Fei-Fei Li: If We Want Machines to Think, We Need to Teach Them to See
  3. Andrew Ng: Why ‘Deep Learning’ Is a Mandate for Humans, Not Just Machines
  4. The Man Behind the Google Brain: Andrew Ng and the Quest for the New AI
  5. Artificial Neural Networks Get a Boost: Computer Chip Replicates Neuron Activity
  6. Machine-Learning Maestro Michael Jordan on the Delusions of Big Data and Other Huge Engineering Efforts
  7. Thinking in Silicon.
  8. Using large-scale brain simulations for machine learning and A.I.
  9. Building artificial nervous system (seem like biophysical modeling)
  10. Neuroscientists Are Making an Artificial Brain for Everyone
Videos
  1. DeepMind Founder (Shane Legg) talk on machine intelligence
  2. Video about Deep learning by Geoffery Hinton
Misc.
  1. Tutorial: Neural Networks, the Human Brain and Learning

Related Posts

  1. Building Machines that Learn and Think Like People (pt 5. Thinking Fast) - January 06, 2018 | machine-learning brain
  2. Building Machines that Learn and Think Like People (pt 4. Learning as Rapid Model-Building) - January 06, 2018 | machine-learning brain
  3. Building Machines that Learn and Think Like People (pt 3. Developmental Software) - January 01, 2018 | machine-learning brain