Chapter
Encouraging Students to Define Their Own Problems for Better Results in AI Projects
23:40 - 33:19 (09:38)
Researchers are advocating for teaching students how to define their own problem and find their own data sets for AI projects instead of using pre-existing data sets, to yield better results in AI projects. This method is ideal for small data sets with around 100 examples.
Clips
This podcast discusses the number of bits the brain needs to learn in a lifetime, with an estimated 10 to the 14 bits needed for synaptic connections and 10 to the 5 bits per second needed to learn in a lifespan.
23:40 - 27:38 (03:57)
Summary
This podcast discusses the number of bits the brain needs to learn in a lifetime, with an estimated 10 to the 14 bits needed for synaptic connections and 10 to the 5 bits per second needed to learn in a lifespan.
ChapterEncouraging Students to Define Their Own Problems for Better Results in AI Projects
Episode#73 – Andrew Ng: Deep Learning, Education, and Real-World AI
PodcastLex Fridman Podcast
The scale and quality of the data set used for training is often overlooked, but increasing these factors can lead to greater breakthroughs in deep learning and surpassing human level performance.
27:38 - 30:24 (02:46)
Summary
The scale and quality of the data set used for training is often overlooked, but increasing these factors can lead to greater breakthroughs in deep learning and surpassing human level performance.
ChapterEncouraging Students to Define Their Own Problems for Better Results in AI Projects
Episode#73 – Andrew Ng: Deep Learning, Education, and Real-World AI
PodcastLex Fridman Podcast
Working with small data sets is much harder than working with big data sets.
30:24 - 33:19 (02:54)
Summary
Working with small data sets is much harder than working with big data sets. Even with small labeling errors, the impact on the data set can be significant.