mostafab

Alibaba Cloud has data centers in Mexico

That's a signal of being SOTA.

Are people here building Frontier AI Models by Distillation? Did any of you try OpenDistil? Is it better than in-house distillation infrastructure?

Why not just an ICO?

Black female scientist Timnit Gebru fired - the end of Google as a top AI research institution?

https://melwy.com/blog/black-female-scientist-timnit-gebru-f...

Ignoring future Google research in ethics has low-risks, as top ethics talent will certainly avoid working for Google, and prefer academic freedom elsewhere (where reviewers are external, to avoid CoI).

Can you post a reference? Some people are doing a lot of ultimatum with me these times

French President Macron seems to be doing what this post is writing: trying to fine-tune the spread, and find a balance between economy and health, with proportional measures

posts from health professionals can be even worse. this one was not too bad at explaining the thinking of ruling elites (who think they can fine-tune an epidemic)

yes,they are better than GAN and reinforcement learning for generating new molecules https://medium.com/the-ai-lab/you-didnt-need-deep-learning-t...

Interesting, I think it can be applied to the current political crisis in Algeria https://medium.com/hirak-tech/alg%C3%A9rie-comment-faire-soi...

PR is what awards are all about

The knowledgeable interviewer should be Schmidhuber, that will be fun

Academic geopolitics. The "deep learning mafia" has striked again, excluding the guy from the other side of the Atlantic, Schmidhuber

Because researchers have little incentives to commercialize their research, especially in academia. They take pride of this insulation from the 'real world'. That's the root cause why research metrics are different from industry metrics.

that's a defeatist attitude. If you had unlimited money, how would you spend your time? that's how you will find what to work on.

Work on AI for chemistry. It has applications to both Energy and Healthcare. Where are you based?

there is nothing new in this paper, and I already wrote about it here: https://medium.com/the-ai-lab/artificial-intelligence-in-dru...

I disagree. There is a cost in publishing bad work, and an incentive for 'predators'. For example, MIT published a bad paper at this ICML, and I was incentivized to tell it publicly: https://medium.com/the-ai-lab/mit-paper-in-ai-for-drug-disco...

I read it and wrote a reply: https://medium.com/@mostafab/mit-paper-in-ai-for-drug-discov...

He can try Startcrowd platform, it is a side-project platform designed with Googlers in mind: https://hackernoon.com/no-kaggle-is-unsuitable-to-study-ai-m...

This professor gave an argument in his paper. I gave a refutation. If you agree with him that char-CNN was a sophisticated model in early 2017, then you are not well-informed about the situation in deep learning.

The MoleculeNet co-author gave another argument in the comments. My refutation is that you can't claim to lack time after 8*20 man-months have passed.

See MoleculeNet ;)

Yes, especially 2)

I am not sure to fully understand your remark, but if you have a benchmark graph convolutions vs. char-CNN, it would be great to write your result and post it on Arxiv. Pande will be interested ;) The problem is not with the theoretical motivation, but with the empirical confirmation.

I also agree that there are many ways to work through the perfect discriminator problem for ORGAN. But it remains to be done (afaik).

From a user viewpoint, Deepchem would greatly benefit from being a better team player with lower-level (Tensorflow) or other (Pytorch) frameworks. The pace of research (in NLP in particular) is too fast to make it realistic to port everything in Deepchem without an unreasonable delay.

Does it fit the Deepchem agenda? That's another question ;)

Sorry for this inconvenience. Content without ads is rarely free.

In general, you are right. But in this particular case, char-CNN is the standard used by many people, Stanford included. It is not sophisticated at all.

Look at the big list of exotic models they tried: http://moleculenet.ai/models

For me, this omission is a negligence. It is selective laziness.

The author (me) does not have 20 PhD students, postdocs and startuppers under his hand, to do the job for Stanford. With my limited resources, it is more cost-effective to do my job against Stanford ;)

Author here

1. I agree that graph convolutions make a lot of sense. However, it is not shown yet that they are better than SMILES, although it might just be a matter of time. A lot of graph variants are possible, or maybe we should look at molecules not as graphs, but as quantum objects (See the other Stanford paper on atom convolutions).

2. Yes, Deepchem is built on top of Tensorflow. This additional layer looks useful, but is it really the case? Using standard NLP models is simpler than using deepchem models, and NLP models might still be the state-of-the-art for chemistry tasks, via SMILES (until we get a reasonably comprehensive benchmark, which MoleculeNet is not).

I am not buying into a chemistry-specific library, until it is shown to be really necessary.

it has been done here: https://arxiv.org/abs/1712.02734