My AI Kick-start Guide
This is my guide of how I took my first steps and developed a full learning plan.
New adventures are amazing, they make your blood flow make your heart beat faster than a cheetah, and gives you goosebumps. During action you feel like the world stopped and there is no such thing as time, all you want is to freeze that moment of joy and adrenaline rush.
That’s how I feel when discovering and learning new Technologies, taking that first step and coding your first “Hello World” which in ML is making a small deep neural network to classify digits from the MNIST dataset. This gave me that shot of adrenaline rush and joy. Some might say it is like an addiction, but ever since I took that first step I couldn’t stop. Instantly fell in love with Tensorflow and Python, it was then that this insatiable curiosity started to grow inside me.
At first, I thought that completing the Tensorflow get started tutorials would be enough. In programming, you just need to know the syntax and coding logic and so I thought machine learning & artificial intelligence worked the same way.
Since google advertises “ You don’t have to be a PhD or be good at math to triumph in the field” all of the low level stuff you should at least know how and why it works is packaged under a function you call(like a black box).
Tensorflow is an open source machine learning framework for high performance numerical computation that offer APIs,a high-level API for beginners which I don’t recommend at all if you don’t have knowledge of machine learning so as a low-level API for experts which I recommend you should know your way around it if you want to build great products levering this amazing alien like Technology.
After messing around with tensorflow I got stuck for the first time and also had this gut feeling that I should know and understand whats under the hood what lies behind those pretty functions like tf.keras.layers and GradientDescentOptimizer. So my curiosity started to grow exponentially and I started to question every pretty function call.
- What is loss ?
- What is Gradient Descent?
- What is softmax ?
- Why and how does this contribute to the prediction ?
After a while I felt demotivate stuck and had no clue where to start digging, so I stopped and gave up. But that was eating me alive “I’m not a quitter!” I thought, with a strong Web Developer background it scared me a bit but I decided to make the change, “What’s the worst it can happen!?”, I thought.
My first solid step
So after months of researching and managing my CSE degree in a foreign country where I have no friends or family near, I discovered something the Andrew Ng Machine learning course which many will disagree but for me it has been a blessing for me, in my point of view there is no better introductory free course to Machine Learning course in the entire MOOC(Massive Open Online Courses).
This post is based on my own experience and what worked for me and might not work for you, there is only one way to find out !
I think every beginner should go through this course to have strong foundations and understanding of the basic math behind it, what is linear & logistic regression, the cost function, gradient descent, what are features and many other basic Machine learning concepts.
This course is not all theory it has some hands-on assignments which are a great way to understand how everything comes together. Like Richard Branson said:
“You don’t learn to walk by following rules. You learn by doing, and by falling over”
Phase II
Immerse myself and dive deep in the field, that meant staying up-to-date with every new concept or model. In a high-level understanding at first only where those concepts and models could be applied and why they were the best and of course as I’m progressing trying to understand how they work via Research Papers. If you find a paper which is interesting you can always implement it or look for the code on github to understand better how to implement it and how the pieces fit together. For the latest and best Research paper visit https://arxiv.org
Check out Sentdex’s youtube channel he is the best when it comes to DIY projects related to AI, the living embodiment of learn by doing. He cuts to the chase and shows you step-by-step how he makes his projects and releases the code .
I also advise you check out Siraj Raval’s youtube channel he is the best WIZARD when it comes to explaining and pushing valuable AI content in the form of videos. I say he is the Gandalf of AI, he is very inspiring and he gives this amazing weekly challenges at the end of his videos and he personally reviews your project’s coed which is an interactive of staying up-to-date. Other way is through twitter if you follow the best and brightest like :
- Andrew Ng, Baidu (@AndrewYNg)
- Amit Singhal, Uber (@theamitsinghal)
- Elon Musk, SpaceX & Tesla (@elonmusk)
- Greg Brockman, OpenAI (@gdb)
Many others are there go check out this post from Li Jiang where he talks about the top 10 people in AI.
Phase III
Before you continue, if you want to take AI seriously maybe as a career path , build a consumers product or deploy it to production, I have some tips before we get into what courses, frameworks and many other resources you need, we have to cover the basics. No matter what high level ML API(Keras) or automatic model creator (Auto) you are using, if you don’t have the following covered you will not evolve fast enough or at all :
- Python syntax understanding;
- Programming logic ;
- OOP(Object Oriented Programming) paradigm;
- Strong ML basics;
With that said, this phase’s objective is to get your hands dirty and make some complex models and dive-in on a subfield of Machine Learning called Deep Learning.
Deep learning is part of a broader family of machine learning methods based on learning data representations, as opposed to task-specific algorithms. Learning can be supervised, semi-supervised or unsupervised.
Deep learning has outperformed every other Machine Learning method and it’s responsible for the latest state-of-the-art Machine Learning models.
I will not list the ML models but if you want to know check-out https://www.kdnuggets.com/2018/03/top-20-deep-learning-papers-2018.html
So how can I get my hands dirty with some AI
There are a lot of resources and it is overwhelming to pick one course or a place to get some projects. I went through more than a dozens sites looking and trying learning resources and come out with three champions namely:
They take a really out-of-the-box approach when it come to teaching Deep Learning. If you have coding skills and high school maths they can teach you with great hands-on projects how to get state-of-the art results in short amount of time.
This is a 5 course Deep Learning specialization from one of the GodFather’s of AI Andrew Ng himself. Hands-on experience designing, building, and deploying end-to-end AI solutions through curated content and instructor-led workshops.
Is a great place to be if you want to learning by doing, the have 9500+ dataset and many kernels for you to play with and learn from so as prize winning competitions for more info checkout this blog post from Nityesh Agarwal entitled Use Kaggle to start (and guide) your ML/ Data Science journey — Why and How.
There are many other resources there I handpicked and pointed out the ones that work for me or worked, feel free to comment down below and share your thoughts.
Its normal to feel stuck and unmotivated but you have the power to change that only you can change your destiny, check-out my blog post about Getting started with AI is Scary !! and I can give you some of the tips I used to overcome every fear of getting started.
“Remember great things take time to open so be resilient and stay hungry.”
If you liked this post please give me a warm clap👏🏽 👏🏽,it will contribute and mean a lot to me, giving more strength to get you the best AI news.
Links and resources:
