LinkedIn or the art of pretending to be an expert.

Long time no write. Been really busy learning some stuff @ku_leuven. Will try to write a couple posts later to finish explaining both my experience in Leuven, and to write a bit of my thesis on Natural Language Inference.

But, not today! Today is about LinkedIn and their inhabitants. LinkedIn, as you know, is a social network that used to be some serious deal, and right now seems more like the playground of recruiters posting their complex mathematical problems that if you solve, you are a genius.

Be a genius, today!

Part of being in social networks, is being in groups, were amazing content is posted and everyone in these groups are experts.

I came across this article. I couldn’t resist to read it (not that I tried hard or anything), as an eager learner that I am. Also the Venn diagram is wonderful, isn’t it? And because I just had my first coffee of the morning, I’m going to use a few minutes commenting on the article.

Before starting, I’m no expert in the matter, so forgive me god/good people of the internet, but I’ll be commenting on the article.

I will be using the text from the post as is, so no misinterpretation can happen.

For as precise a profession as we data scientists purport to be we are sometimes way too casual with our language. Read several articles about AI, Deep Learning, and Machine learning and you will come away confused whether these are all the same or all different. Imagine how confused non-data scientists must be.

Rock solid start. I mean, I’m no expert but, Quora sometimes has some questions that perfectly answered, and I have found some very well written explanations. Btw, shoutout to the Q&A sessions from Quora, latest I read from @fchollet, some interesting questions, with really interesting answers. (I think this is the first time in this post that I’m being serious). But I know I know, Quora is quite new in the internet, and stackoverflow doesn’t solve these tricky questions, let’s check wikipedia. I know wikipedia is not the way to go but, here is another link.

THIS IS FROM WIKIPEDIA NOT THE ARTICLE.
Deep learning (also known as deep structured learning, hierarchical learning or deep machine learning) is a branch of machine learning …

To read a bit more about it, found really useful this. But some insight about the discussion can be found in this NVIDIA post, you know, the guys who build the GPUs so we can play videogames, errr train deep networks. I really meant train deep networks. Let’s move again to the article.

The truth is that each of these terms has some overlap in a Venn diagram but none of these is a perfect subset of the other and none completely explains the others. Let us explain.

Well, I’m sure I gave some explanation about it, but it’s worth the check the venn diagram, because diagrams are fun.

By the way, it is amazing that the author of the post, cites himself as a source when he ellaborates on his points. Not sure who did he cite on his first post, or if an index out of bounds error arised. (Sorry. Not a funny joke.)

The article proceeds, the author does not explain any AI algorithm (not explaining as in a scientific way, but any example, anything), doesn’t even bother to mention approaches based on logic or anything. A little bit of Turing tests, a little bit of weak vs strong AI and a bit of robots. (I’m still amazed that no external sources were used to generate the article). To end the AI part, the author mentions that there have been some significant advances in AI, increasing performance in some areas.

From http://blog.dilbert.com/

Next part he writes about Deep Learning (DL), and says, all of the previous advances are coming from DL, (the advances he mentioned in the AI section). Then he proceeds to say that the advances in DL are because of the advances on Neural Nets models, and because we now have better chips in which run these algorithms. Kinda forgetting that we can make use of this algorithms because data collection has been significantly improved in all aspects.

These is becoming super long and boring for the reader, so you get my point, lets check the final paragraph, Machine Learning. The author says that there are two definitions about it. First.

In the new usage, machine learning is defined to mean only algorithms used in unsupervised pattern identification. This is intended to be essentially the same thing as Deep Learning leaving aside the inconvenient fact that Deep Learning still needs to be shown examples of correct output, making it not truly ‘unsupervised’.

I mean. What to say. I’m certainly no expert on deep learning, if anything I know, it’s about supervised learning, but this is about using google (an important skill of any data scientist out there, really, use google properly). Takes about 10 segs if you are clicking the search button, to type “unsupervised deep learning” and then click the button.

Screenshot from Google results 22/08/2016 12.30

Yea, the fact that some deep learning algorithms need labeled data, in any Machine Learning setup, is called Supervised Learning. When data is not labeled, it is called unsupervised learning. Second definition:

In the more traditional usage still widely used including by me, machine learning means the application of any computer-enabled algorithm that can be applied against a data set to find a pattern in the data. This encompasses basically all types of data science algorithms, supervised, unsupervised, segmentation, classification, or regression.

Really surprised that cloud computing and big data are not types of data science algorithms. I’m assuming, the author refers as segmentation (clustering) as unsupervised learning algorithms, and classification and regression as supervised learning algorithms, a definition that extends in any kind of setups, not matter if its machine learning, deep learning, or real human learning.

The cherry on the top.

It’s clear that the traditional usage would be much broader than Deep Learning but would not encompass manually assembled expert systems in AI.
Next time you’re reading about AI, Deep Learning, and Machine Learning make your understanding more precise and more valuable by applying these explanations.

Thanks teacher.

I don’t know, I’ve been talking about LinkedIn and “experts” for so long with collegues/friends that I guess today I felt like writing about it. It takes more time to write the post, and look for nice venn diagrams, than to actually gather the knowledge about certain topics. I know LinkedIn is all about selling oneself, and I’m not the best at it, but in science, any kind of science, please, take five minutes of your life to write something coherent, something that can be used by others to learn.

Here you have another Venn diagram. Because they are truly amazing. Have a nice day.