jorgemf

Stars mean different things to different people. Some people use them to like something, others just to bookmark it for later.

  Location: Spain
  Remote: Yes
  Willing to relocate: No
  Technologies: Deep Learning, TensorFlow, Python, kotlin, java
  Résumé/CV: https://jorgemf.github.io/cv.html
  Email: https://jorgemf.github.io

Last 3 years working for a SF startup in deep learning for object detection in images and other problems related with images. Most of my job is to apply latest research papers ideas to solve problems in the company.

Side projects in NLP with deep learning.

Working in AI since 15 years ago.

You would want the performance for new deep learning applications (for example better images with low light conditions)

yesterday was too late to make backups...

I think you just pointed the biggest issues in the article: the problem starts before any of the solutions proposed. You can do whatever you want during the job/university but if you still get 95% males it won't fix the issue.

> I think we have to start early and stop dividing professions into for men or women. Let men be nurses and women firefighters. Look at your childrens personalities and let them grow in whatever they enjoy, girly or not.

I would just add that we should do this even if it doesn't change anything. Just let people do whatever they like and don't judge them with stupid stereotypes. Don't force girls to be CS and boys to be nurses if they don't want to.

One machine with 8 V100 GPUs. If you consider one TPU pod a single machine the TPU is faster. Those numbers also show that 8 GPUs are slower than 8 TPUs (so same conclusion as the article)

Each country in Europe has its own visa, and depending on the visa you might need other visas for other countries if you want to travel around. So the first question is where do you want to go? Do you know other languages than english? (north countries speak English either as first or second language) Oh, don't forget one country is leaving the European Union in the next years (it might impact you).

First find a couple of places you want to live in and then search for their visas, doing it the other way is going to be a nightmare as you already know.

The best way is to have a girl/boyfriend that only speaks english. The second best way is to move to a country where they speak english and avoid people who is not native. So basically the way to becoming fluent in english is to use it as much as you can. Movies, books also help, but for communication you need to talk a lot.

That service doesn't justify the price tag for me.

145 per user per month, not 25. Not all countries pay 10k a month to their employees

Video games are a great environment because they are quite complex and they are already coded and we understand them. For basic research it doesn't matter whether is it a game, test cases or 3d design. They use games because it is less work as they already work and are easy to integrante. Moreover each game is a different domain.

I would like to use it but I think the price doesn't justify the tool. For a team of 5 people github is $25 a month, you are $745 a month. I do understand a price a bit higher that github but not 30 times more expensive.

Regarding your points:

1) "explainable AI" is better in GA than in deep learning. GA gives you a structure that works and probably easier to understand than any Deep Learning model (which it is a huge math function). There are so many things that also doesn't make sense why they work in deep learning but we still use them, that is the same with GA.

2) Knowing the fitness function doesn't mean you can solve the problem. When the search space is so big you need something to search on it, and there is where GA can shine. It is also the same with Deep Learning and mostly evolutionary computation. The search space is so huge for a "brute force algorithm". You need heuristics and GA works well for some problems, the same way gradient descent works well for others too.

an OEM of a technology that noone wants to use?

I hope the content is really good and worth the experience, otherwise this is useless. But I am still afraid that if they have to create their own content it might not make sense as a business, why people wouldn't be interested in created content for the next audiovisual platform?

I don't think the PC is the problem here.

I think this is just incredible. Let's suppose they have an incredible technology and that works (which it is not true at this time because they don't have any product out yet). How do they justify the product adoption which such high valuation? (with patents of products they also don't have any idea if the people will want them?).

Even it is the best technology ever, if not enough people buy it the company will go down very quickly.

exactly, the someFactor is the error of the next layer calculated with the derivades of the functions (as you would calculate the minimum of a funtion using the derivades). The tricky part is to calculate the derivades, but since auto differentiation we can do a lot of cool stuff.

Backpropagation is based on linear optimization (aka calculate the maximum or minimum of a function based on the derivades of that function, this is taught before university in my country). And also in the chain rule to calculate the derivades of functions (first year of university).

But I meant, if you see the equations and the steps without understanding completely the insights, it is a joke of an algorithm. It just does some multiplications and applies the new gradients, move to the previous layer and repeat.

I have read a lot of papers and usually, you end up in the original one if you check the references. But you have to check papers in specific areas. I am not aware of any good document where everything is there.

> his first impulse was to write backprop from scratch

backprop is a very simple algorithm, nothing to fear there. The problems are to calculate the derivates if you want to be flexible building your model. But for feedforward networks with sigmoid activation, the equations to update the weights are a joke.

> not any sort of fundamental algorithmic paradigm shift

I think I cannot agree with this. There has been a lot of improvements to the algorithms to solve problems and the pace has speed up thanks to GPUs. You just cannot make a neural network from 15 years ago bigger and think it is going to work with modern GPUs, it is not going to work at all. Moreover, new techniques have appeared to solve other type of problems.

I am talking about things like batch normalization, RELUs, LSTMs or GANs. Yes, neural networks still use gradient descent, but there are people working on other algorithms now and they seem to work but they are just less efficient.

> This is similar to how the integrated circuit enabled the personal computing "revolution" but down at the transistor level, it's still using the same principles of digital logic since the 1940s.

This claim has exactly the same problem as before. You can also say evolution has done nothing because the same principles that are in people they were there with the dinosaurs and even with the first cells. We are just a lot more cells than before.

> "you can't use AI/ML unless you have a PhD/5 years research experience"

This is not true since a few years ago. But the fact that you can use it doesn't mean you understand what is happening and why it works in development but not in production. Everybody can copy a jupyter notebook and train a TensorFlow model in ImageNet. Now go to a new domain with very few information like 3D models and create a new network to be trained in that dataset. How many people that can train ImageNet can do the latter? Even inside deep learning experts in image classification fail in reinforcement learning domains and need a couple of years to be completely productive.

You can use it with GPUs: https://towardsdatascience.com/fast-ai-lesson-1-on-google-co...

I hope the target audience is not researchers. I think it doesn't make sense for researchers because they are usually very focused on a topic and don't care (or don't have time) about papers on other topics.

I see this as a tool to increase the hype about Machine Learning. Only people starting in the topic will use it.

It will make sense to filter the whole list with keywords or topics or anything that filters the papers and shows only the ones related to your researcher. The subs still have too many papers.

0.9 wasn't production ready and they didn't guarantee backward compatibility until 1.0. So nothing to change from 1.4 to 1.5, maybe you will have some warnings about features that will change in the future, but it will work.

Speed in FP16 under k80 is almost as FP32, the architecture doesn't work well with FP16. As you said, you need Volta to notice improvements with FP16.

The other two main features are: Eager execution and TensorFlow Lite

for the same logic the researcher is another hyperparameter

(I know you are right, but so many people here think AutoML is exactly the same that the HPO they were doing since long time ago)

Does it really say it somewhere? As far as I know when you train TensorFlow models they are stored in gs, I thought this would be similar. Otherwise I don't know how they are planning to integrate this with the api they have to upload your models and make requests.

HPO means Hyper Parameter Optimization? Because AutoML has nothing to do with it, AutoML is mostly about the architecture of the model, not about Hyper Parameter Optimization.

90% of the time companies apply regression to problems because is the tool they know and works well for a lot of problems. But I think that has changed with deep learning. Deep learning works well for a lot of complex problems (as images). So now companies can solve problems that before couldn't solve. That is why I think it is relevant to see the trend.