How To Self Study AI FAST

before anybody makes a comment yes I do know that my hair is wet but I got to go somewhere after this video is for my short attention span friends who still want to learn AI so usually when you're trying to learn something new it look something like a straight line first you learn calculus linear algebra probability statistics programming machine learning deep learning Etc so kind of like that progression don't get me wrong you do need to learn these things eventually but my problem is that I can't even get past one of these subjects without getting really bored getting stuck and giving up hey no hate these are amazing resources so what if it's not the resources themselves that's the problem but the way that we use them is there a way that we can learn Ai and not give up introducing the renon method or if for some strange reason you don't like Naruto the concentric Circle method how does it work so in the middle of the renegon you have the thing that you want to learn which is AI we go from the middle and we go outwards so for the small circle we just need to learn the basics of AI such as a high level of how machine Learning Works how large language models work but most importantly how do you use these models with python don't worry I'll go into a lot more detail about exactly what you need to learn and recommend some resources later but the point is that you learn just enough so that you're able to build a really cool AI thing like this study tool or personal AI assistant as soon as possible like I'm saying realistically 1 month if you have zero coding experience and a week or two if you have some intermediate experience in Python and then after you do this we take this excitement and satisfaction of building this really cool thing and we use that as motivation to go into the next layer of the circle we dive a little bit deeper into what exactly is machine learning how does it work as some of the math surrounding it which would allow us to then build something else that is really cool and then use that as motivation to expand again into the next level of the circle so you kind of just repeat this cycle so that you're learning more and more advanced things and also getting to apply them until you become quite Advent and be able to tr truly understand AI models like how chat GPT works and even build your own so what is machine learning let's start with a hot dog do pizza yes do pizza that's that's it it only does hot dogs no and a naha dog so that was an example of machine learning machine learning is a way for computers to learn and make decisions by themselves by studying a recognizing patterns in data there are many different types of machine learning models and this one specifically is called a CNS a convolutional neuron Network by the way I might be throwing some terminology here and there but don't worry about remembering things and understanding I'm just putting these here so as you're learning you kind of go like oh like she talked about this like I'm learning this right now I'll be explaining more about how these work later in the video as well but first let's talk about how Jimmy was able to build this hot dog not hot dog model so first you have your little baby model that has not seen the world yet and you got to start feeding it images about hot dogs but you also have to show it pictures of not hot dogs you also want to show it some tricky cases like this dog that looks like a hot dog and this hot dog sausage doesn't have a bun I don't know if that's still considered a hot dog is that actually a hot dog though anyways you do this many many times and it starts to learn what is considered a hot dog and what is not considered a hot dog or more specifically what are the features that make it more hot dog like and what are the features that make it less likely to be a hot dog for example if it sees this cylindrical reddish thing it makes a note that this is an increased likelihood of that being a hot dog and it sees this white stuff around this red thing and again it will make a note that there's an increase in a likelihood of being this hot dog it will then come up a score with its prediction of How likely it is a hot dog but for example if it sees this triangular looking thing it goes like huh triangles are not hot dog like so it decreases the likelihood of that being a hot dog and so on and so forth until it gets better and better at predicting whether it's a hot dog or not a hot dog now let's take a look at Chachi BT over here which is also a machine learning model except in this case the data we're feeding it is a bunch of text Data like the entire internet's Text data and it uses his data to predict the next words in a sentence it's based upon its previous words for example if you have the words I am and the word sleeping it'll give a likelihood of that being the next word which is probably relatively high but there can also be a word like potato which probably has a pretty low likelihood of being an next word so the algorithm picks the word with the highest probability and it somehow magically is able to chain these together to form coherent sentences isn't that crazy like thinking about how it actually works of course I'm simplifying things a little bit here for now though what's very exciting is that you can actually use these AI models pretty easy easily to start building your own AI products say like this AI personal assistant that's able to schedule your life and stuff and by easily I mean if you have zero knowledge about coding it'll probably take you about a month or if you have some intermediate level of coding it'll take you like less than a week or two what you need to learn first is the basics of python variables data types if statements Loops objectoriented programming and apis which stands for application programming interfaces and it's for interacting with other people's software I'm also going to give you some suggestions for resources brilliant has a super beginner friendly course which is super interactive which is great for people with very short attention spans cuz you can like you know do the little dios and click things and things pop up you can get started with brilliant for free they also are the sponsor of today's video If you prefer video learning there is this really good introduction to python from free code camp and if you're into text or reading textbooks so me personally I'm not that into textbooks because it makes me really bored easily but I have heard that this book automate the boring stuff is a really good introduction I want you to especially focus on understanding apis and how to use them because that's how we're going to be able to use these AI models that other people made next we're going to learn the very Basics about large language models which are the AI models that power chat Bots like chat gbt brilliant also has a crash course on large language models which is super beginner friendly like you don't even need to know how to code but if you're into videos this is a 1-hour introduction to large language models by Andre karthy who is an expert in this field next up we're going to do this course on prompt engineering for developers this course is only an hour long and is completely free from deeplearning.ai but seriously this is such a good course in starting to build AI products using open AI apis it teaches you prompt engineering en able to interact with AI models and how to connect and use the API to access the models all right at this point you have the basics of building AI products you can use open AI apis in order to build chat Bots and personal assistance you can also generate images from Models like do also link some more apis that you can use to generate text to video and other cool things you can do also link some examples of projects that you can build Link description you now know how to use AI models through apis but you still don't really know how they work or how to make your own to be able to do that it's kind of like building the foundations of the building you need to lay a very solid foundation first by getting a better understanding of machine learning but before we can dive into the machine learning algorithms themselves we still need to take a step back and break it down into its sub fields of fundamental mathematics statistics and programming specifically in Python what you need to learn one at this intermediate level of python you need to start learning more modules that are related to data manipulation because you need to use data in order to teach your machine learning model stuff so we need to learn the modules of numpy pandas matpa live for data visualization and pyit learn for building machine learning models there are so many great tutorials and courses out there and I'll link them below free code Camp is probably my favorite resource and if you're into books python for data analysis I've heard is very good now math these scary stuff A lot of people are intimidated by math I am also intimidated by math math so the good news is that you don't need to learn that much of math you don't need to sit there and learn how to do like derivatives by hand you just need to understand like the concept of calculus the contract of what a matrix is for linear algebra how to use probability to determine the likelihood of something that's about to happen these are the foundations of machine learning models for my short attention span friends especially I feel like for math math is like especially challenging because it's it can be so boring brilliant is nice and interactive and it gives examples of things so I recommend the coures calculus fundamentals introduction to linear algebra and introduction to probability you can also take this math for ML specialization free on corsera if you want to dive a little bit deeper next up statistics you got to know things like descriptive statistics inferential statistics hypothesis testing Central limit theorem distributions confidence intervals it sounds like a lot but it's pretty much just first year statistics in college again brilliant is how I personally brushed up and learned more about statistics but I also love supplementing with my all-time favorite techdata YouTuber Josh starmer He is very short attention span friendly because how can you possibly get bored of someone singing about math don't be afraid of neur networks they're not scary if you want something more thorough there's a Standford course on corsera called introduction to statistics by the way a pro tip especially for subjects like math that are kind of like conceptual and hard to understand using chat PT as a personal tutor is literally a game changer it can help explain difficult Concepts and give analogies for things uh where you can like use it to dive deeper into stuff so I'm not going to go into too much detail about how to do that because I already made a video which I'll link over here talking about how to use CH PT as a learning tool highly recommend all right all right now we have truly laid a very solid foundation and we can now dive into machine learning yay machine learning as a field is very very large and there's a lot of different aspects of it so I only want you to focus on understanding the categories of different algorithms and some of the example algorithms out there stuff like regressions K means clustering decision trees Etc and understand the difference between supervised and unsupervised learning Josh starmer is absolutely my go-to for machine learning content I give Josh full credit for me actually graduating my Master's Degree because I took with this really hard machine learning course and yeah like I would not have graduated without him if you wanted something a little bit more thorough there's also the Stanford and deeplearning.ai course called machine learning specialization all right we' have expanded into the next [Music] Circle neurons are cells in your brain that form a network so that you're able to think and do stuff now ai is modeled after our brains we have these nodes that represent neurons which create what we call artificial neuron networks if you feed these neuron networks data it's able to start learning by itself kind of like when a baby is first born it doesn't really like have anything in its brain but as it starts having more experiences collecting more data it's able to start learning by itself now if you start stacking layers and layers of these neurons together things start getting really interesting and you can create models that are capable of doing incredible tasks this is called Deep learning cuz you got a lot of layers stack together and it's like very deep it's an advanced subfield of machine learning the hot dog no hot dog am model is a model that uses deep learning specifically in the field of computer vision and the AI models that powers chat Bots like chat gbt are called large language models they also use deep learning in the field of natural language processing okay so at this point we're another layer deep and learning about deep learning deep learning layer deep and learning about specializations like computer vision and large language models some recommended resources Brilliance introduction in neuron Network covers the basics and the artificial neuron Network course goes into deep learning again Josh sarmer is just is the best and if you want to go a little bit deeper there's a corsera specialization in deep learning now at this point you can also start branching out into different sub Fields like for example you're interested in hot dogs and not hot dogs you can dive deeper into computer vision and here's also a free corsera specialization on it or if you're interested in large language models and things like that you can dive deeper into natural language processing here's another specialization of corsera a final quick tip okay I do know I give a lot of different resources here but that's mostly just to give you guys a variety based upon what kind of learning style you have do not I repeat do not try to go through all the different resources and try to like learn every single little thing and get like really obsessed with everything just choose one of these resources they're all amazing go through it and then start building your own projects you can build your own neuro networks contribute towards open source AI models and fine-tune other people's models now I want to talk a little bit more about the sponsor of today's video brilliant thank you brilliant I've already mentioned them a few times especially for short attention span friends because they're so interactive brilliant actually only specializ izes and stem subjects so that they're able to make the best courses to teach these subjects I personally love using brilliant whenever I want to learn new things and brush up on different skills especially the math and stats part I get so bored when I try to just like watch a video or do some courses um so just you know having those like little interactive things helps a lot in my understanding they have Timeless course offerings like math and stats programming and python as well as new course offerings like the neuron networks courses for the Deep learning and introduction to large language models you can join a millions of people already learning on brilliant I head on over to this link to get started for free also linked in description if you go through my link the first 200 people will get 20% off in annual membership all right that is the end of today's video thank you guys all so much for watching let me know in the comments if you're now interested in learning Ai and if you want me to make more videos related to learning AI things I don't know if you guys are into that okay anyways have a wonderful day and I'll see you guys in the next video or live stream

Channel: Tina Huang

Views: 466,077

Keywords:

Id: 0zi-CG_iB6w

Length: 12min 54sec (774 seconds)

Published: Sat Dec 30 2023