Machine Learning with Thermodynamic RAM and the Knowm API

Intelligence is our ally for understanding the world around us, and at Knowm Inc. we want to build the most intelligent machines the world has ever seen. We have taken our inspiration by observing examples of intelligent structures in nature and noticing that natural systems compute efficiently by combining computation and memory into the same space, effectively removing the barriers current von-Neumann computer architectures face. So far we’ve demonstrated through simulations that breaking these restrictions are possible if we build intelligent machines which have access to a neuro-memristive synaptic processor we called Thermodynamic RAM (kT-RAM). The emerging theory of AHaH Computing, which forms the underlying basis of kT-RAM is explained in depth here.

Neural Network

Image by speedoflife

Machine Learning Series

This series of articles/tutorials is intended to summarize the capabilities we’ve so far demonstrated with the Knowm API and familiarize you with some basic Machine Learning applications and benchmarks. It starts with an introduction to Machine Learning and then moves on to kT-RAM, the Knowm API, and then finally Classification, Clustering, Prediction, and Robotic Actualization. These articles are an in depth extension on what we already published in 2014 in our PLoS paper: AHaH Computing–From Metastable Switches to Attractors to Machine Learning. Each category has a number of examples which you can use to gain a better understanding of that technique. If you are familiar with the basics of the Knowm API and Machine Learning then feel free to jump forward to the specific tutorials, otherwise I would recommend progressing linearly. This series is a work in progress. To see the newest articles as they are published, subscribe to our RSS Feed. We look forward to hearing your comments or questions and invite you to join in discussions at The articles contained within scratch the surface of coding with the Knowm API. Full code for all benchmarks, additional benchmarks and applications as well as in-depth tutorials for understanding the essentials of AHaH Computing and kT-RAM are available for all Knowm Developer Community members.

Table of Contents

  1. Introduction to Machine Learning
  2. Crash Course in kT-RAM
  3. Linear Classifier with the Knowm API
  4. Primary and Secondary Performance Metrics
    1. Census Income Classification Benchmark
    2. Wisconsin Breast Cancer Classification Benchmark
    3. Reuters 21578 Classification Benchmark
    4. ‎MNIST Hand Written Digits Classification Benchmark
  5. Introduction to Clustering
  6. Clustering with the Knowm API
    1. Clustering KNN Encodings
  7. Signal Prediction
    1. Complex Sine Wave Prediction
  8. Reinforcement Learning
    1. Robotic Arm Motor Control

Further Reading

TOC: Table of Contents
Next: Introduction to Machine Learning

Related Posts

Subscribe To Our Newsletter

Join our low volume mailing list to receive the latest news and updates from our team.

Leave a Comment


Subscribe to our low-volume mailing list to receive important updates and announcements directly in your inbox.