Gold Blog, Aug 2017A New Beginning to Deep Learning

I won't give you the clichéd line that it's never too late because that's not the point. It is actually because, a term that I loved as soon as I came across it- 'The AI Winter' - doesn't seem to ever be going to return again.

Most of you guys out there must be thinking... What’s exactly going on around me! What is AI? What is Machine Learning? Am I too late to this party?! And, wasn’t all this enough that people threw in a new term called deep learning? Great, another thing I have absolutely no idea about!

Calm down. You're not late at all. I won't give you the clichéd line that it's never too late because that's not the point. It is actually because, a term that I loved as soon as I came across it- 'The AI Winter' - doesn't seem to ever be going to return again.

It’s not for the first time that the field of AI has gained popularity. It has had its own share of rapid ups and downs in the past, usually followed by a downfall in funding and interest, often referred to as the ‘AI Winter’. The first winter occurred in the 1970s, followed by another one in 1980s for some reason or the other, but majorly due to less resources. I agree that there have been many major breakthroughs but here’s my attempt to illustrate the timeline of major events...


So, how has everything suddenly fallen in place this time around?

What’s changed in the recent times is that researchers and scientists have finally opened the treasure trove of both, the enormous computational power and the abundant storehouses of data in the form of images, video, audio, and text files scattered across the Internet which are the only resources that they were not having with them.

Data scientists and AI researchers are currently paid amazingly well as companies build research teams hoping that they'll get unexpected breakthroughs. Most of them might not even be knowing what exactly to expect from the person they're hiring considering how new this field still is, but they seem to go with the flow and hype in the industry created for AI.

Andrew Ng, the man behind the famous ML course on Coursera that each one of us must have heard of, said in an interview in 2016:

"There’s definitely hype, but I think there’s such a strong underlying driver of real value that it won’t crash like it did in previous years."

AI is here to stay and it's going to get bigger and bigger by the minute. The hype isn't for nothing nor is it temporary like the previous ones. AI won't go away and we all need to accept it sooner or later.

Although the other quite hyped winter is coming, but the AI winter is definitely not coming anytime soon, maybe even never again! So, you're not late at all, even if you're reading this months after it was actually posted. I'm sorry but I can't say the same if you're reading this after one year or so considering how rapidly this field is progressing.

Yoshua Bengio, a pioneer in the field of AI said:

"Most people do not realize how primitive the systems we build are, and unfortunately, many journalists (and some scientists) propagate a fear of AI which is completely out of proportion with reality. We would be baffled if we could build machines that would have the intelligence of a mouse in the near future, but we are far even from that. Yet, these algorithms already have very useful technological applications, and more will come. That being said, I do believe that humans will one day build machines that will be as intelligent as humans in most respects."


But, what exactly are we trying to achieve out of all this fuss?

It’s just a very simple idea. Imagine the basic magical things that you do everyday that you never think highly of. You just don’t think too much as to why can you 'see' things, 'hear' things, 'say' things, 'understand' things, whereas things can’t do the same. Computers started off as objects that made our work a bit easier by taking the calculation work on their shoulders. Now we’re just trying to put more of our work on its head because, why not? Its shoulders are already busy with our basic calculation work but there’s space on its head too! So why not, right?

We're doing this by making it as similar to us as possible. We're all lazy asses, seriously! But quite ironically, we're working really hard to make that happen. So I guess we're not too lazy if you look at it that way.

Fei Fei Li, Chief Scientist of AI/ML at Google Cloud and Director of Stanford AI and Vision Lab called the present times as the "the beginning to transform the 4th Industrial Revolution", quoted one of her favourite lines at the recent famous all-female discussion panel at Google I/O '17:

"There are no independent machine values. Machine values are human values."

We're trying to make machines see, hear, understand, learn and most importantly, learn from previous mistakes. That's something we all are told to do and we know we should do but many of us still don't.

Deep Learning is worthy of all the credit for the latest and daily advances in computer vision, speech recognition, natural language processing, audio recognition and what not! You name it and deep learning has a solution to it almost everytime.


Okay... so why am I here?

The main purpose of this blog is to make the most important way to stay afloat in this vast AI sea available and easy to comprehend for everyone. The way I’m talking about is through research papers. Yes, those highly dreaded PDFs that get downloaded on their own once you click on some paper you found on maybe and then hell breaks out! You were advised by people to read research papers but little did you know what was in store for you. You can’t follow most of what’s written in there because obviously it needs you to be knowing a lot of other things.

Basically, it's like you need some context. You can't start off with some TV series from the 7th season, right? Eventually, you give up just after a couple of paragraphs into the paper, and it's natural! It happens so many times with you, me and just everyone out there. My aim here is to give that context and make research papers or rather concepts easy to understand.

I prefer giving you the context in short and in the best way possible instead of just providing you some random links, leaving you to drown in yet another bunch of information you might not have any context for. Research papers are undoubtedly great sources to find out about the bleeding edge technologies that have been developed in AI although here I shall be concentrating more on deep learning, but we need to understand it to get something good out of it.

I don't even know to how many eyes would these words of mine reach but those to whom they reach, I want my posts to be as helpful and informative as possible. That's because I'd love to share as much as I can from whatever I have learnt till now and will continue to keep learning, so that you have an easier time fighting to learn what seems to have the capacity to eat up most of the currently existing manual jobs. Mind you, this revolution has already been set rolling!

So, rather than being deserted helpless when the time comes for you to be replaced by a machine, be on the other side making machines and have a goddamn job! But I must say, this logic is far from enough to survive in this field because you need immense determination and more importantly passion, to keep up with the new approaches to older developments that pop up almost every single day.

Elon Musk has been extremely vocal about his fear of Artificial Intelligence:

"Humans must merge with machines or become irrelevant in AI age."

I hope I haven't scared you too much but I had to, a little bit atleast. That's because AI is the need of the hour and you've got to get equipped with this new weapon unless you want to be replaced by a machine in the near future. Beware, because it's nearer than you imagine it to be!

I wish to develop an in-depth understanding of concepts in the minds of readers of this blog by always keeping the mood light. This is because I’ve realised by observing as well as closely following the works of Youtube ML rockstars Siraj Raval and Harrison Kinsley (@sentdex) that the best way to connect with people willing to learn is to think like them and try to solve every problem they might face while going through your content not by being serious with a straight face on, but instead enjoy the teaching and learning process.

I wanted to keep the first post like an introductory one, trying to update you with what’s going on around you because I know it’s really tough watching AI get so much attention from every direction and everyone wanting to be a part of this much hyped topic yet not being able to figure out how and where to start!

I’m not very different from you. Even I’m as confused and anxious about these sudden developments around us as you are. It’s just that I embarked on this journey quite early. Now, after having been in this field for some time and having learnt with time as to what to do and what not to do, I simply wish to help you guys out there who are as confused and as messed up as I was over a year ago.

In my next post, we’ll be diving into the basic terminology that every single deep learning enthusiast shall keep coming across wherever anything about deep learning is mentioned, how are AI, ML and Deep Learning different from each other, and a lot more. You’ll also get introduced to basic feed forward neural networks which is the first step towards being able to make your own Artificial Neural Network. Until then, stay tuned!

I’ll be sharing some interesting resources and sites at the end of every post. It’ll be great for you all to get into the habit of reading blogs and articles on various sites because the quality of content you’ll find would always be inspiring as well as informative.

These guys are life savers! Check out their channels for ML and DL tutorial videos:

Few Articles you may find interesting:

Bio: Raksham Pandey is a Data Scientist in the making... Electrified by AI... Passion for Deep Learning to solve problems that matter.

Original. Reposted with permission.