Building Machine Learning Models with Clojure and Cortex
summary
Cortex is an open-source Clojure ML library built on a CPU/GPU compute backend with CUDA support, representing networks as maps so a net as Clojure maps (nodes edges buffers streams) is fully inspectable and mutable. Data pipelines use lazy sequences for infinite batching and simple CSV → numeric features → one-hot labels preprocessing. Workflow: write a minimal layer description, call build to get the graph, then train with train-network. A credit-card fraud model reaches ~F1 0.8 on a dataset with ~0.1% positives.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:43What is Machine Learning
- 7:49How Machine Learning Works
- 13:45Why Clojure
- 14:54Function Composition
- 18:22Data Processing
- 21:20Cortex
- 25:21Neural Network Design
- 30:37Data Loading Preprocessing
- 36:44Additional Model Tuning
related talks
25mMore Clojure Less Complication
44mMachine Learning with Artificial Neural Networks in Clojure
32mBrain Data in Clojure
41mDeploying ML models in a Clojure environment
26mThe Feynman Machine: AI architecture in Clojure/Neanderthal
28mMachine Learning Live
18mClojure Versus the Botnets
27mReal World AI Integration