jorgemf

In the last model they trained they shared some part of a layer for all the agents. So it was like a one agent trained with shared knowledge of the rest.

It is very hard to know exactly what the client wants. Also it is possible the client will change its idea in the middle of the project. You basically use your experience to have an idea of how much will take you to develop the project and then add a safe margin. Where this safe margin usually is multiply per 2 or 3 of your initial estimation. You should also consider how big/complex is the company you are going to work for, the bigger the more difficult is to get what you need from them. So you multiply again per 2 or 3 if the company is big to account for the delays in getting what you need from them and other bureaucracy stuff.

The other option is to charge per hour with a range of hours you will work on the project. And if they want to make changes or add things you just charge more hours.

In my country is illegal to have someone as a freelance doing the same as a full time job. So I see the point of upwork. If that company is your only client and you are working full time with them, they should hire you (at least in my country). In my country you have more rights if you are full time employee than if you are a freelance, as for example a monthly pay after they fire you to help you to find another job

You are not alone. With time series that is the easiest solution for any model. That is why you need a lot of work creating your datasets and the experiment setup with time series. To avoid the easy solution and obtain something useful.

This information is for outside and inside the company. When you interview someone you cannot go with a predefined answer that you want the candidate to achieve.

Practice, practice, practice. There is no other way to improve at anything. If you want to be the best, you have to practice as much as you can. Probably will take years or even your whole life to by the best or one of the best. Nothing comes easy in life.

Probably Nvidia is not interesting in this type of tools and this is more like "hey, look what you can do with a lot of Nvidia cards, buy us a lot of them and you will do this and much more"

> But for analytics and prediction it’s been extremely useless compared to actual people.

Usually the problem is that companies are going to sell you an AI solution for your data, but these companies are not going to tell you that with your current data there is very little an AI system can do. I say this because I saw companies who wanted a prediction system with 2 years of data (40 datapoints per year). Any AI expert will see there is no way this will work but still the system is sold to the company. This is much worse in the public sector.

> Resources are optimized relative to their current values

I see that as the problem. We have limited resources and don't care very much about the future of them. Current value vs long term value.

Free resources shouldn't be an excuse to use them wisely.

I get your point, but it is not a good reason why we cannot do things better.

Why do you think your churn rate is high? Do you know ow what is the average churn rate of other apps or similar apps? (No, your churn rate is not that bad, it is normal in apps). Another good question is the life time value of each customer and how long your users stay in your app (like since they installed it until they uninstalled it). Create an engagement metric for user use case and try to I prove it, more engage means less churn.

Focus on your target market and measure the churn rate in your target market. Forget the rest of the users, those are noise. Develop what you need for your target market.

Neural networks are based on human brains. So, how are they both different? It is a very hard question, because if we would understand better the brain we will simulate it better in the computers, how do we know we are not already doing a good simulation of the brain? Any answer to this question would be a step forward to improve our artificial intelligence systems.

Archlinux user here (Manjaro is based on Archlinux). TensorFlow with GPU support is a supported package you can install it with all it's dependencies in a simple command with Pacman. But you don't want to do that the same way you don't want to compile TensorFlow. It is better to use an official docker image because they include all the dependencies, even the optionals as nccl 2.0. You forgot to install it, and it is speed up if you use more than one GPU. Just use the official docker image and forget all the burdensome of configuring everything.

P.S. well you need to install docker with Nvidia support, but you can use it in the future for other things. This is also a couple of packages in Pacman (maybe Nvidia docker is in air, I don't remember now).

Well, there are many things to consider:

- Code is small, most of it is calling well-know libraries with the algorithms.

- Every 3-4 lines of code you usually show the results of what is happening (either how the data changed, a graph or whatever)

- 95% of the errors in data science come from bad practices: you either didn't clean up the data correctly, you didn't split the train/test/validation sets correctly or at the right step in the process, you chosed the wrong algorithm, etc

- most of the time you don't know whether you are doing something wrong or not because you don't have enough knowledge, but the code shows something all the time. No bugs in the code.

Those are the reason I don't care very much about bugs in the code because usually there are bigger problems in the data science process rather than in the code.

IT depends what you want to do

Jupyter notebooks are for data science, mostly because visualization is required. Something you do once, report it and it is done. Itsn't make sense to use Jupyter for other stuff. It doesn't make much sense to use them for training big models of deep learning because there are better tools for that.

How big are your datasets?

How do you use Jupyter for Deep Learning? I can only think about prototyping stuff, but not for training a large dataset (that doesn't fit in memory) for days.

What's the difference between the output of a network and what an expert says? Unless you can probe mathematically why networks or human mind works, there is no explanation for any of both methods. I can argue that humans learnt from examples the same ways neural networks do, you can say the opposite. But we have no way to say that that any claim is false or true.

So as I said you can copy what others do. That is fine, but you don't k ow deep learning, you know how to apply it based on examples, which is is fine for a lot of things.

I wasn't talking about back propagation. But sometimes you need to change the loss function, or the shape of the network. Or combine two models. Back propagation is the same for those examples, but not other math stuff.

The only thing that makes you think it is easy is because you are just copying what others have been doing and you don't change anything. Try to go beyond that and you will change your mind quite quickly.

Because most objects are made for human hands. So it is better to have a robot with human-like hands (or body) than to change the shape of all the things we have already created. So future robots can use our human-made stuff.

You cannot learn machine learning or deep learning in a few months. You can learn to copy what these guides do, but if you want to do something slightly different you will feel you know nothing (because you actually probably don't know anything about the maths behind why the things works, so when you want to change them you don't know how)

Deep learning is far from slow down. The research speed has increased a lot in the last 5 years. It amazes me how fast the researchers are able to come up with new things. From one year to the next one there is something shiny and new.

I don't follow very much blockchain but seems like it was 6 years ago

It sounds to me they want people to learn about deep learning and tensorflow. In the long term there will be more professionals and the salaries will go down. That is how they will make profit from this type of thing.

We do have imitation Learning. I think the article is missing some important parts in RL. One way to train a network is to use experience from others or even past experience of the agent, but why is it interesting doing it from scratch? Because doing it from scratch will allow us to face a lot more problems where we don't have any information or skills and because we avoid the bias in the data and we can discover new things (as it happened with AlphaGo zero with its 'tactics' in go). Now we have a method that can be applied to all the problems that are similar to a board game without any other information, just the rules of the domain.

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

I am a remote Machine Learning Researcher/Engineer for different startups of San Francisco. I have an MBA and I have also pursued a PhD in Artificial Intelligence. I was graduate with first class honors and I have been awarded several by the research and work done, both locally and internationally. I love maths since I was very young and I have been finalist in s couple international math contest I joined in by hobby.

My goals in life are being happy, unveil the secrets of the human mind and create Artificial Intelligence thanks to maths.

> Training a model in the browser is the least practical use for tensorflow.js

Unless you want to do some type of federated training in real time without sending private information to the servers (very unusual case in my opinion)

This is all the reasons why I want to use your tool.

Thanks for creating it and I hope one day soon to contribute to your project.

I don't know pachyderm but it seems to me quite similar to storm to create data pipelines. Polyaxon is useful to train deep learning models in a cluster. I couldn't find any example of how to do it in pachyderm (there are examples with only one node).