The most efficient route to become a developer on SingularityNet

development
study
beginner

#1

Dear community,
I also want to get involved in AI development on SingularityNet. I just saw the code that Ben posted on the forum and I found it very complex. I am not a very technical person but I programmed in Perl in the past and I am a very hard worker. Are there any tips available on what i can study that would enable me to get into developing for SingularityNet specifically?
I do understand that it might be very, very difficult and arduous to get to a level that would enable me to contribute to this platform in terms of development.
I am now studying python in combination with AI concepts but I want to spend my time as effectively as possible. If python would be optional for example, then maybe I would be better off spending my time taking another direction. I know Aigents will make it easier to develop for people that are not already very skilled in AI development, but I do not know how that will work yet. Are there any tips in what I can study now to be able to participate later?


#2

If you are learning Python and AI concepts, you are already well on your way to becoming an AI developer. That’s actually what I would’ve suggested to you even if you didn’t mention it, as a lot of AI tools and frameworks nowadays use Python. It’s also a very nice language to learn. :slight_smile:

After that, you can choose what field of AI you’d like to pursue. The most dominant field today is probably deep learning (it’s what all the big companies are using for their services). There are other approaches to AI, of course, but deep learning has achieved state-of-the-art results in a number of fields.

Personally I’d recommend taking an online course on the subject. Among the most popular ones seem to be the Udacity courses (they recently launched this School of AI):

And Andrew Ng’s deep learning course, if you want to take a deep dive into deep learning:

At the end of the day, you have a lot of resources available online, so seeing that you said that you’re a hard worker, I don’t doubt you’ll be a proficient AI developer in no time. :slight_smile:

After that you can start with SNET-specific development, of which I leave the advices to someone with more insight into the matter. But you should be able to import your previous AI models into SNET, so if you learn to create deep learning models, you should be good to go when the platform launches.


#3

Generous and good advice Simon. Good on you mate!


#4

Dr. Ben offers a few thoughts on this here: AGI Curriculum


#5

One way to really get into machine learning is to find an interesting problem you can solve with the existing frameworks.

I few years back I’d been away from the machine learning world doing prosaic system architecture and backend development. I read and followed the examples in this article to train my own facial landmark detector, incrementally getting better performance as it introduces various deep learning practices:

http://danielnouri.org/notes/2014/12/17/using-convolutional-neural-nets-to-detect-facial-keypoints-tutorial/

If you’re like me, then trying to just learn algorithms can get very boring, but applying them to specific problems I find motivating. It gets me to focus on what I need to learn to improve the solution step by step.


#6

Thank you for your answers, much appreciated:) I will definitely check them out.


#7

The quickest, cheapest way to learn AI programming with python and build a github portfolio is to read these three books, in this order:

  1. https://www.amazon.com/Make-Your-Own-Neural-Network-ebook/dp/B01EER4Z4G
  2. https://www.amazon.com/Introduction-Math-Neural-Networks-Heaton-ebook/dp/B00845UQL6/ref=sr_1_2?s=digital-text&ie=UTF8&qid=1526278965&sr=1-2&keywords=ai+neural+network+math
  3. https://www.amazon.com/Hands-Machine-Learning-Scikit-Learn-TensorFlow/dp/1491962291/ref=pd_bxgy_14_img_2?_encoding=UTF8&pd_rd_i=1491962291&pd_rd_r=5GM0YQW93CJ4NYZSCVM3&pd_rd_w=w28Xu&pd_rd_wg=eqGVl&psc=1&refRID=5GM0YQW93CJ4NYZSCVM3