*Function* draws from a dropout neural network. This new visualisation technique depicts the distribution over *functions* rather than the predictive distribution (see demo [below](http://mlg.eng.cam.ac.uk/yarin/blog_2248.html#demo)).

This is Yarin Gal's PhD Thesis. It is awesome for people who want to learn Bayesian Deep Learning.

The thesis can be obtained as a Single PDF (9.1M), or as individual chapters (since the single file is fairly large):

• Contents 目錄(PDF, 36K)
• Chapter 1: The Importance of Knowing What We Don't Know 知所不知的重要性(PDF, 393K)
• Chapter 2: The Language of Uncertainty 不確定性的語(yǔ)言 (PDF, 136K)
• Chapter 3: Bayesian Deep Learning 貝葉斯深度學(xué)習(xí) (PDF, 302K)
• Chapter 4: Uncertainty Quality 不確定性的質(zhì)量 (PDF, 2.9M)
• Chapter 5: Applications 應(yīng)用 (PDF, 648K)
• Chapter 6: Deep Insights 深層洞察力 (PDF, 939K)
• Chapter 7: Future Research 未來(lái)研究(PDF, 28K)
• Bibliography (PDF, 72K)
• Appendix A: KL condition KL 條件 (PDF, 71K)
• Appendix B: Figures (PDF, 2M)
• Appendix C: Spike and slab prior KL (PDF, 28K)