Chapter
Importance of Foundations in Machine Learning
50:02 - 58:43 (08:41)
The speaker encourages deep learning enthusiasts not to skip the foundational learning of machine learning as it can limit innovation and result in incorrect applications, especially in complex systems like medical diagnosis.
Clips
The speaker stresses the importance of having foundational knowledge when learning machine learning to allow for innovation, problem customization and recognition of artifacts in the model.
50:02 - 55:21 (05:18)
Summary
The speaker stresses the importance of having foundational knowledge when learning machine learning to allow for innovation, problem customization and recognition of artifacts in the model. He shares that he was initially surprised that neural networks work so well for end-to-end learning, given their large search space contrary to previous assumptions.
ChapterImportance of Foundations in Machine Learning
Episode#93 – Daphne Koller: Biomedicine and Machine Learning
PodcastLex Fridman Podcast
The issue with machine learning models is that they can provide an incorrect diagnosis or solution with complete confidence, especially when encountering an outlier in a data set.
55:22 - 58:43 (03:21)
Summary
The issue with machine learning models is that they can provide an incorrect diagnosis or solution with complete confidence, especially when encountering an outlier in a data set. The ability for the network to identify and communicate its uncertainty is crucial in mission-critical applications like medical diagnosis and automated driving.