Date post: | 10-Feb-2017 |
Category: |
Leadership & Management |
Upload: | francis-pieraut |
View: | 62 times |
Download: | 1 times |
Let me explain...● How do you train a Neural Network (Machine Learning Algorithm)?● You compute the gradient (gradient = expected outcome - outcome)● The bigger is the gradient, the faster your Neural Network will learn something● Human by nature are lazy which means they follow the lowest gradient. Why?
○ Before: error = you could get killed -> no reward for errors, the stick. ○ Our reward system culture is still “the stick as opposed to the carrot”. Error is bad and you get punished instead of get
rewarded to learn. ○ Laziness -> People want to minimize their energy consumption (scarcity of food, now abundance).
● So basically, retrain yourself to follow the hardest gradient. ● Put yourself in the hardest situation possible and reward yourself when you learn from errors.
Lazy people Slow learner Ultimate learner
Same principle phrased differently● Get out of your comfort zone● Fail fast● Minimal Viable Project● Do it● Agile● Get a mentor● Jump● ...