Making machine learning accessible using Keras
summary
Keras is a high‑level Python neural‑network API that runs on TensorFlow, CNTK or Theano and claims about 300,000 users with ~754 GitHub contributors; it emphasizes simple, consistent APIs like one-line fit and model.save/model.load, and says the Sequential model sequential covers 70% of use cases. TensorFlow integration adds TF.data pipelines, eager execution for imperative debugging, and DistributionStrategy for multi‑GPU (and evolving TPU) training; supports functional models, subclassing, custom losses and metrics.
AI-written summary · about
chapters (26)
- 0:00Introduction
- 0:55Outline
- 1:12Machine Learning
- 2:26Deep Learning
- 2:43What is Keras
- 3:55Why should you use Keras
- 4:50Who writes Keras
- 5:19Brief history
- 5:45Sample companies
- 6:50Model Building
- 7:51Sequential Model
- 8:38Training and Saving
- 10:33Training
- 11:22Data Validation
- 13:28Saving
- 15:03Functional API
- 16:13Functional API Example
- 17:57Model Subclasses
- 18:44Tensorflow Integration
- 20:26Estimator
- 21:02Functional Model
- 21:27Gigger Execution
- 21:50TensorFlow Distribution Strategy
- 22:39Timeline
- 23:24Conclusion
- 23:42Questions
