Fei-Fei Lee, The doyenne of Deep Learning
There were many contributions of women in the history of Computer Science
There seems to be many great female engineers. Sorry I only talked about men so far.
In a field of machine learning leading the trend, there is also a female engineer of unrivaled ability.
Her name is Fei-Fei Li.
Fei-Fei Li and her family moved from China to the United States when she was 16, and achieve the “American Dream”.
Her parents were intellectuals back in China, but earned their living by fixing cameras and working as cashers in America.
Fei-Fei Li herself raised tuitions by purchasing the laundry machines and running the cleaner’s.
After graduating from Princeton University, she declined job offers from McKinsey and Goldman Sachs,
continued her study, and became the doctor of Artificial Intelligence.
Fei-Fei Li became known to the world by IMAGENET.
In 2012, the SuperVision team led by Geoffrey Hinton won an overwhelming victory in the IMAGENET contest,
with the Deep Learning technology.
IMAGENET might be the reason this Deep Learning technology was able to reappear to the world.
This Deep Learning developed year after year, and in 2017,
it eventually exceeded human’s capacity by reaching the accuracy rate of 97%.
Fei-Fei Li, the creator of IMAGENET, argued that in the process of learning, the learning volume is far more important than the algorithm itself,
based on the learning process babies go through, thousands of learning before they can recognize the object.
Let’s hear from Fei-Fei Li, the creator of IMAGENET.
Thus, she made a plan to secure a huge amount of image data for learning.
Meanwhile, some colleagues advised her to work on more useful materials to become a tenured professor,
The research fund for completing the data base was not enough as well.
At first, she payed the students in graduate school $10 an hour, and led them to categorize the materials.
Of course, that would have costed a lot of money.
Nevertheless, there is always a path if you never stop dreaming.
She heard about a crowdsourcing website called MTurk.
With MTurk, she could finish the IMAGENET project in 2 and a half years.
For one’s information, MTurk is a website opened in 2006, which provides services
that connects the survey requester and respondents.
It costs only 10 cents for 10 questions, so it enables people
to accumulate a large number of survey responses.
Like this, IMAGENET, a massive warehouse of images with 15,000,000 labeled images,
in 22,000 categories, made by 48,940 people from 167 countries.
Recently, she is making an open course about computer vision and Deep Learning with her pupil, Andrej Karpathy.
Fei-Fei Li’s ultimate goal is to make machines have the eyes of human.
Then, she believes, it will give a huge benefit to the world, and to people.
What do you think?
She deserves the name, “Doyenne of Deep Learning” don’t you think?