[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math


View post   

File: 152 KB, 2048x1448, Machinelearning.jpg [View same] [iqdb] [saucenao] [google]
11934793 No.11934793 [Reply] [Original]

Hey /sci I am from /g--- don't bully me pls. I have to work on machine learning, natural language processing and data science for my new company starting November. Looking around I found lot of books and resource but they are mostly python and do it and move on. Problem is I want to understand the theory behind those domains but I have no idea where to start?
What is neural network?
How can it be mathematically represented?
What is natural language?
Pls help.

>> No.11934867

>>11934793
https://www.youtube.com/playlist?list=PLZHQObOWTQDNU6R1_67000Dx_ZCJB-3pi

>> No.11934872

>>11934793
https://en.wikipedia.org/wiki/Natural_language

>> No.11934900

>>11934867
>>11934872
Book source?

>> No.11934901
File: 211 KB, 941x1080, 1576016698515.jpg [View same] [iqdb] [saucenao] [google]
11934901

>>11934793
Well you see. There were still incels in the 90s. Instead of sperging out they gathered in forums and, as expected of an obsessive incel, began discussing and mapping out human behavior in hopes of solving it, manipulating it and eventually getting sex. Not under the form of a sanitized journal on psychology but rather in an unrestricted discussion devoid of morals much alike that of 4chan's. The central theme of that manipulation was called NLP. Neuro-linguistic programming. It was the idea that by uttering specific words in a specific sequence, that you could make anyone fall for anything without even noticing it. Basically this https://www.youtube.com/watch?v=Yr-QtNE9k84 .
However the material got buried under the ruins of Web 1.0 and it is no longer accessible no matter how hard you dig for it. They have all hoarded it inside their brains and have transformed it into what they now call "marketing", becoming quite successful thanks to this. About 90% of the OGs are millionaires and the majority of them are pioneers in modern social media marketing, mostly multi-millionaires.

Me personally? Well I was too young when all of this happened but I still got on the train on the very last minute, around 2006, and learned many things while they were still accessible. I am now re-mapping the entire theory about them and are redesigning them for imageboards, which are the future of humanity's cultural development as text is the least metabolically-expensive information transmission method, and will use them to form a new theory on imageboard behavior so that I can steer the discussion towards what I perceive is beneficial for humanity in my own subjective world view. Not just 4chan but whatever appears after it as well. I am about halfway done with ironing it out and will have a deterministic set of reproducible results within the next 5 years, before the neural networks get to there first.

Hopefully this answers your question OP.

>> No.11934932

>>11934793
>What is neural network
An acyclic directed graph where each node takes in a vector of inputs (received from nodes pointing to it) and spits out a vector of outputs (sent to nodes it points to).
The function at each node is what the training is is trying find.
>natural language
Sloppy, imperfect language.
This can be due to poor grammar, poor pronunciation, noise, etc.

>> No.11934934

>>11934793
Check gen.lib.rus.ec for books

>> No.11934936

>>11934793
Bishop (2006) is a good theoretical intro book that covers a bunch of machine learning stuff. Some of it is outdated at this point, but it still serves as a good guide to most of the history. I'm not sure about NLP books in particular.

>> No.11934952

>>11934901
She is susceptible. Is there a scale for where people fall on suggestion and hypnosis?

>> No.11935004

>>11934793
I have a BSc and an MSc in AI. The field has changed very quickly over the last ten years. Neural networks are almost the only serious game in town now. It’s good that you want to gain a deeper understanding, but the fact is that we don’t really have a deep understanding, in part because it has changed so quickly. The AI field has become almost entirely empirical, meaning we just throw stuff at the wall and see what sticks. My advice would be to just try to implement some neural networks using PyTorch or similar, which make it relatively easy nowadays. That way you’ll get some intuition of what works and what doesn’t.

>> No.11936426

>>11934901
NLP is fairly interesting. Are you willing to publish any of your notes in pdf? Even if it’s schizo rambling I’d consider trying some of it on my rooommates, Derren’s videos were cool