Gan's father, Ian Goodfellow, resigned from Apple: I don't want to go back to the office

Ian Goodfellow

Author | Wang Yue

Edit | Chen Caixian

According to a tweet by Zo Schiffer, author of The Verge, Ian Goodfellow, Apple's director of machine learning and father of Gan, will leave Apple. In a note to employees, Ian wrote, "I insist that flexible working is the best policy for my team. （I believe strongly that more flexibility would have been the best policy for my team）"。

Picture note: Foreign media reporter tweeted the news

In this way, part of the reason Ian left Apple (and probably even a large part) was because of dissatisfaction with the resumption of work policy.

It turned out that according to Apple's recent announcement of the resumption of work policy, employees need to return to the office from April 11. At first, you only need to go back to the office one day a week, and the rest of the time you work from home, but as time goes on, employees gradually increase the time they spend back in the office. Work in the office for at least 2 days by 2 May and 3 days in the office after 23 May.

For employees who are accustomed to telecommuting during the pandemic and have found a balance between telecommuting and life, it is understandable that they will not adapt to the intensive rework arrangement. In addition, considering the commuting problem and the possibility of contracting the virus, how the company arranges to resume work has become an important issue in the post-epidemic era.

1

Ian Goodfellow its people

Ian Goodfellow, who studied at Stanford University with his undergraduate and master's degrees, was mentored by the famous Ng Enda, and at the doctoral level, he studied machine learning at the University of Montreal under Yoshua Bengio, one of the "troikas of deep learning".

Ian Goodfellow

He was considered a top expert in the field of artificial intelligence at a young age because he proposed generative adversarial networks (GANs), known as the father of GAN. As one of the hottest deep learning models now, GAN has become the most out-of-the-loop discussion topic in the machine learning community in recent years.

Inspired by Ian Goodfellow's genius idea after drinking, GAN is inherently topical:

The GAN consists of two models, a generative model and a discriminative model. The task of generating a model is to generate instances that look natural and realistic and are similar to the original data. The task of the discriminant model is to determine whether a given instance appears to be naturally real or artificially forged.

This left-right mutual competition has gradually evolved into an increasingly powerful data fraud ability, which not only opens the door to a new world, but also raises many ethical questions. For example, foreign netizens "deepfakes" replace the faces of celebrities and celebrities into indecent videos, or use the faces of public figures to fake news, so as to achieve the purpose of spreading rumors or causing public opinion.

As too powerful a being, GAN is like a double-edged sword, whether it is retracted or released, it can always attract the attention of the world.

Ian Goodfellow's Google Scholar page

The picture above is Ian's Google Scholar page, which can be seen that his paper has more than 180,000 citations.

And gan alone is the first, with more than 40,000 citations.

In second place, Deep Learning, an AI foundation book launched by Ian in collaboration with his phD directors Yoshua Bengio and Aaron Courville, is now considered a must-read for machine learning.

Note: Deep Learning entity books

With so many dazzling achievements, this time Ian left Apple due to the resumption of work policy, and what is hidden under the dash may be the future he has long planned.

Let's wait and see.

Reference Links:

https://www.leiphone.com/category/ai/iL1S8jkc4ytZFzHS.html

https://arxiv.org/abs/1406.2661

https://twitter.com/zoeschiffer?s=11&t=GHcyU6vqd4K7HXGk8z4tUg

Leifeng Network