Machine Learning Exposed Deep and Reinforcement Learning
summary
Gentle walkthrough of neural nets that actually builds and runs examples: a Java web app (Java + Spring + Angular 2) visualizes feed‑forward math on XOR and the iris dataset, using sigmoid/tanh activations and a softmax output (softmax outputs add to 1). Training demos use gradient descent/backprop with tunable hyperparameters (seed, iterations, learning rate, hidden neurons — examples use 4 hidden units and mention scaling to ~260), show cost falling (≈1.0 to ≈0.1 after ~600 iterations), and a Kaggle speed‑dating model giving ~81% probability for attractive=9 (TensorFlow layers and matrix/vector math are recommended).
AI-written summary · about
chapters (17)
- 0:00Introduction and speakers
- 1:27Machine learning overview
- 2:44Core learning paradigms
- 3:44Neural network architecture
- 5:30Application demonstration
- 7:42The intuition of neural nets
- 10:48Network structure explained
- 14:06Logic and feed-forward
- 15:29Feed-forward mathematics
- 17:32Refining the network model
- 20:15Vectorization and libraries
- 23:20Bias and layer dynamics
- 25:23Optimization considerations
- 30:16Backpropagation explained
- 32:50Gradient descent nuances
- 37:25Real-world dataset demo
- 44:21Training results and Q&A
related talks
44mMachine Learning Exposed
50mMachine Learning Exposed: The Fundamentals
49mMachine Learning Exposed: Introduction to Machine Learning
43m(Deep) Learn You a Neural Net for Great Good!
26mMachine Learning in the Browser with Deep Neural Networks
51mGetting Started with Deep Learning for Software Developers
01h 50mNeural Networks: Walkthrough
37mCode First, Math Later: Learning Neural Nets Through Implementation and Examples