Patterns, Predictions and Programming
summary
Describes a concrete Java ML stack: JSR 381 Java ML API with a Deep Nets reference implementation and an Amazon implementation; demo trains a visual classifier on 872 Chihuahua images, yields ~88% on similar photos but only 66% on a mushroom image, showing underfitting and need for more data or longer training (minutes versus hours). Covers using curl and openai-java clients against ~70 GPT models, notes GPT knowledge cutoff Sep 2021, and cites studies showing ~50% productivity gains when pairing with models.
AI-written summary · about
chapters (23)
- 0:00Introduction and speaker background
- 1:04Machine learning as a tech revolution
- 2:26History and ubiquity of patterns
- 6:19Design patterns and regular expressions
- 8:47Machine learning fundamentals
- 10:32Deep learning and generative AI
- 12:40Need for Java ML APIs
- 17:11JSR 381 and visual recognition
- 18:35ML workflow and data preparation
- 20:00Demo and classifier training
- 22:28Introduction to ChatGPT
- 23:10Language models explained
- 25:38Transformer architecture breakthrough
- 27:48ChatGPT API and services
- 32:22Working with ChatGPT API
- 36:16Programming with ChatGPT
- 41:15Model languages and code generation
- 43:44Code review and limitations
- 46:13Advanced ChatGPT use cases
- 49:57Prompt engineering importance
- 51:18Future of ML and desktop Transformers
- 53:23Ethical concerns and job impact
- 55:29Q&A session
related talks
47mML in Java, YES it's possible!
46mML in Java, YES it's possible!
50mML in Java, YES it's possible!
55mMachine Learning and AI in Java on AWS
27mDeveloping and Deploying ML Models with Java
44mDeep Learning in Production
12mPreview: ML Serving
43mWhy Java and the JVM Will Dominate the Future of Machine Learning, AI, and Big Data