New Machine Learning Phenomenon Discovered : Grokking Machine Learning is a field where advancements are occuring every day and apparently scientist Eric Jang has found something! Eric Jang or as he is known on Twitter: (@ericjang11) points out a new and possibly very important machine learning phenomenon called “grokking”.
What exactly is this phenomenon called “grokking”?
In summary, if you just keep training neural networks well past the point of overfitting, a sudden “phase change” can occur improving generalization performance from random chance level to perfect generalization. More data accelerates the speed at which this phase change occurs.
Reproductions of the paper here:
What does Grokking mean for the progression of Machine Learning devlopment and research. Moreover, what is its meaning for the whole Machine Learning world?
We will have to wait to find out!
As a result of the study of computer algorithms that can improve automatically through experience and by the use of data, we have ML. Furthermore, it is a part of artificial intelligence.
Firstly, Machine learning algorithms build a model based on sample data, known as training data. Secondly, this is in order to make predictions or decisions without programmers input. In conclusion, Machine learning algorithms apply to a wide variety of both business and academic applications. Lastly fields such as in medicine, email filtering, speech recognition, and computer vision, where it is difficult or unfeasible to develop conventional algorithms to perform the needed tasks.