#20 – Active Machine Learning with Dr. Jennifer Prendki
In Algo Podcast Episode 20, Dr. Jennifer Prendki and Amjad Hussain discuss Active Machine Learning.
Skip, recap, or review with timestamps:
3:29 Jennifer’s AI Journey
5:15 Collecting Good Data
6:04 Particle Accelerator & Physics Research
8:00 Physics Standard Models
10:18 Are Standard Models Still Useful?
13:49 Data Outliers
15:31 Next Physics Breakthrough
19:12 Aliens, Dark Matter & Energy
21:12 Difference between Antimatter & Dark Matter
24:24 Decay Curve for Electrons & Positrons
26:48 Discovering New Particles
28:32 Relationship Between Quantum Nature & Space Time
30:30 What is Artificial Intelligence & Machine Learning?
32:58 Why can people perform better than Supervised Machine Learning?
37:24 Machine Teaching
38:51 How to improve Supervised Learning
43:02 Focusing on variance instead of volume
43:45 ROI for organizations
45:40 How to build diverse data sets
48:25 Understandability in models & data
50:09 Active Machine Learning
53:33 Understanding where models are struggling
56:09 How to learn active learning approach
58:44 Using Active Learning to Prioritize Data
1:04:08 Role of synthetic data
1:09:41 How can people interact with trained models?
1:12:20 Meta Machine Learning
1:13:09 Advice for people learning ML