The Importance of Problem-Solving in Machine Learning

The technique that's extremely important if you want to really apply machine learning to real-world problems is problem-solving. You might learn all the techniques, but if you don't apply them to real-world problems, it's useless. Tackle calm is a terrific website that has competitions for machine learning scientists and research engineers and data scientists. It's a great source where you can get a lot of datasets, real-world datasets. I suggest you pick a bunch of problems from tackle calm and solve these problems end-to-end. Solve them from how to pre-process data to trading multiple models and understanding why a technique or a model is working or not working.

Understanding the Why Behind Machine Learning Models

Sometimes people just become coding monkeys, trying one model after another without understanding why a particular technique is working or not working. It's essential to answer those questions either using data visualization or using mathematics. For every problem that you solve, please document and code up things and put it on a GitHub or write a blog about it. This will help you build a portfolio, just like artists and photographers have a portfolio of their work. I strongly recommend anyone who is looking to move into a career in machine learning to build a portfolio of at least five plus case studies.

Building a Portfolio for Machine Learning Careers

Most people underestimate the importance of building a portfolio when it comes to getting hired as a machine learning professional. A portfolio is essential because it showcases your skills and ability to solve real-world problems. Without a portfolio, you're just a faceless candidate in a sea of applications. By solving problems from tackle calm and documenting your work, you can create a portfolio that will help you stand out from the competition.

Learning Machine Learning through Practice

Learning machine learning is not hard if you try to break down the mathematics into diagrams, if you try to make equations in mathematics like poetry, geometry like art. If you're willing to put in the effort of 5-10 hours a week for 3-6 months, I think that's all it takes. The only prerequisite for learning machine learning is effort.

Applying Day Course on Machine Learning

I teach a very applied and practical course on machine learning called apply day. It covers all topics from what is the equation of a line to deep learning techniques. We have detailed contents on our website, including free videos. Our course has no prerequisites at all, so anyone can learn it. If you scroll down, you'll get an exhaustive list of all the topics that we cover.

Case Studies and Real-World Applications

One of the most important aspects of machine learning is solving real-world problems end-to-end. That's why I've designed our course to include 10+ case studies that demonstrate how to solve a problem from start to finish. We work with all of our students to build a portfolio with at least five case studies because that's what they can showcase to move into careers in machine learning and AI.

Customer Service at Applied A

As I've learned from my experience at Amazon, customer service is essential when it comes to online courses. If there is no customer service, customers may get stuck or students might get stuck at a concept and not understand why. At applied A course comm, we provide both phone service and an email service where we try to answer customer questions as soon as possible. We aspire to provide terrific customer service because our goal is to help you succeed in machine learning.

Free Videos and Feedback

We have a bunch of free videos on our website at applied a course comm. If you go to our website and click on the check out the free videos, you'll get all these free videos. We would love to hear your feedback on what you think about our course and all the very best.

"WEBVTTKind: captionsLanguage: enso a question that is often asked is how to learn machine learning and AI and how much time does it take and how to actually what is the strategy to learn and my answer here is based on my experience as an instructor at apply day a course comm where we teach machine learning and applied and artificial religions from a very applied standpoint it's also based on my experience as a senior much learning scientist at Amazon my experience at Yahoo labs and startup that I founded where an I'm entered I meant a lot of youngsters in in moving to machine learning careers and in learning advanced techniques in machine learning and AI and the basic strategy to learn machine learning is this you need to learn both theory and practical aspects you can't just say I will learn only theory all I've known or I'll only learn some api's and learn how to program things if you just learn some aps and learn the practical aspects and if you don't know theory you'll always be lacking it's always important to get a very very good intuition on each concept not just the theory not just equations but also to understand each concept intuitively if needed through geometry I always say this I always say that equations in mathematics are like poetry and geometric intuition or geometry is like art right so if you can understand the concept both using equations which is like poetry and geometry which is like art your your intuitive grasp of things because we humans are very very visual creatures and we often gain better intuition when we try to visualize things using geometry the most important of all whenever you learn a technique in any area whether it's probability whether it's statistics whether it's machine learning anything trying to practice practice practice try to apply the technique lots of places to understand where it works where it doesn't work and for every technique you understand try to understand what are the boundary cases what happens in the boundary cases what are the failure cases where does a given technique fail and from my own experience of teaching lots of students machine learning in AI believe machine learning is not very hard lot of people panic when they see mathematics but remember mathematics is beautiful we can make it extremely simple using geometry if you use because V has visual creatures if you try to understand lot of people panic when they see big equations but if you try to read equations in English it becomes like poetry if you try to understand geometry it's like art you can appreciate mathematics better visually and also three equations right I believe I believe that learning machine learning and AI is not at all hard if you are willing to put in the effort the effort is something that we cannot run away from I believe if you can put in five to ten hours of effort of e you can learn machine learning in three to six months time of course there will be some fast learners who can do it faster there will be some slower learners who will take more time but on average we have seen that if you spend ten hours a week you can learn it in three to four months similarly if you put in five hours a week you can learn it in six to seven months and this is based on experience of some of the students that that we have taught over the last few months now this is this is a list of topics and I will give you some references which will help you so the first topic of course is you need to know a programming language preferably Python or R so Python is a very general purpose language R is a very statistical language my personal preference has always been with Python because it's much more easy to write code for those of us who know some programming language like C C++ Python should be very very easy to pick up there are these two very nice books there's a book called learning Python by Mark loads which gives you the overview of the language itself everything from what is a class what is a variable what is the dictionary etcetera - there is a book there's a very specific book by aurélie called Python for data analysis web by wes mckinney this does mostly data analysis term in this you learn specialized libraries like numpy site by matplotlib etcetera etc which are very very useful for machine learning and data science and in general artificial regions the second very important topic is probability and statistics some of you may panic as soon as you see propagate statistics but it need not be hard there is a very nice book called thing stats by ellen downey i believe this is also an overly textbook where where where the wherein the author takes a real-world example and tries to introduce all the concepts in probability and statistics using real-world data by you by asking very simple questions about data it's a very very interesting book especially for beginners and there is an online course called statistics and probability by Khan Academy Khan Academy is one of my favorite resources on the Internet because it because the explanation of concepts on Khan Academy is terrific some of you may realize that what I'm using the interface I'm using is also somewhat like Khan Academy not exactly like Khan Academy and I've learned a lot from Khan Academy over the years the third most important mathematical prerequisite is linear algebra for linear algebra you can easily connect linear algebra to geometry so understanding linear algebra concepts like what is the equation of a plane like what is the distance of a point from a plane we can understand all of that using geometry and the two great sources there is an online course on Khan Academy called linear algebra which is very interesting also there is a very nice book and there's also an online course called introduction to linear algebra by Gilbert Strang Gilbert Strang is a professor at MIT and this book is very very theoretical but it gives you much more mathematical rigor and depth right so but I don't know how much a practical aspect you might learn from this from this book but the linear algebra by Khan Academy is a very very but practical and theoretical it gives you a very well-rounded understanding of the concept now the fourth important thing is calculus and numerical optimization for this for calculus itself you need to know basics of calculus you don't have to know how to solve partial differential equations and advanced concepts basics of calculus what is differentiation what is partial differentiation these simple concepts are good enough and there is a course called multivariable calculus at Khan Academy which is which is a very good course and if you want if you prefer a book or a or a more rigorous mathematical course there is a online course called convex optimization there's also book by the same name by professor at Stanford called Stephen Boyd the first few chapters in this gives you a very very intuitive understanding on how optimization works and what Wow and there is both the book they're also very nice slides for this when I learned optimization myself in my grad school I learned a lot from Stephan Boyd's book and now after we have learned all these basic concepts we move on to machine learning we have a bunch of techniques called classification and regression techniques we also have techniques called clustering techniques and matrix factorization techniques and recommender systems for all of these sources for all of these topics there is a very popular online course called machine learning by andrew inc at Coursera it's a very popular course and there is also a very popular book called pattern recognition and machine learning by Christopher Bishop this book is very rigorous in its mathematically very deep well the online course is very introductory course which with some mathematical detail not too much and some applicative detail again not too much of applicative detail it tries to balance out both of them to some extent and this this book the pattern recognition and machine learning book by Christopher Bishop is a very good introductory textbook which is used in lot of colleges and universities so once you learn classification regression technique clustering techniques matrix factorization and recommendation system techniques you also have to learn some dimension reduction techniques there are a couple of dimensional reduction techniques which are very important to understand data then comes one of the hottest areas in machine learning today called neural networks and deep learning and for this there is a very popular course again online called deep learning by and ruing at Coursera which is very popular I also loved this book this book is one of my favorite books that I've read in the recent past it's called deep learning by an good fellow a very good book I've read this book ever since it was in preprint I really enjoyed reading this book it's it's a it's a it's a heavy book but it's worth reading it if you have bandwidth and for all of these techniques in addition to all these resources the most important thing to learn for each technique learn the mathematics learn the geometric intuition if there is a geometric intuition understand what are the assumptions that each of these techniques is making for it to work properly understand try to draw geometrically if you can or mathematically what is the best and worst case for each of these algorithms try to understand interpretability of each model interpretability for some techniques could be feature importance try to see how you can interpret a model don't ever look at models as black boxes because that way you'll never gain a deeper understanding understand how outliers or noisy points can affect your model understand how bias data sets or skew data sets are are handled in each of these modeling techniques also if you have time please code up each of these algorithms on your own that gives you a very very deep understanding of how each algorithm works for each algorithm there will be minor to major variations learning at least some of the variations will give you a better understanding on how on what what modifications can be made for a given algorithm so similarly for every technique it's extremely important to understand the limitations and how to work it on those limitations lot of people know a bunch of techniques but they don't know where not to apply the technique that's extremely important if you want to really apply machine learning to real-world problems the most important part is problem solving you might learn all the techniques or not but if you don't apply all of those techniques to real-world problems it's useless so tackle calm is a terrific website which which have which has competitions for machine learning scientists and research engineers and data scientists it's a great source where you can get a lot of datasets real world datasets I suggest you pick a bunch of problems from with a cool calm and solve these problems end to end solve it from how to pre-process data to trading multiple models and understanding this is extremely important understand why a technique or a model is working or not working because sometimes people just become coding monkeys we just try one model of the other model after a third model that's why you're not getting the intuition of why a model is working or not working or one why one model is better than other model try to answer those questions either using data visualization or using mathematics and for every problem that you solve please document and code up things and put it on a github or write a blog about it I strongly recommend anybody who is looking to move into a career in machine learning to build a port for you build a portfolio just like artists and photographers and models build a portfolio of at least five plus case studies and put your code explain your analysis draw plots explain why your technique is working why your technique is not working do a thorough analysis of each of these problems and write it it's it's useful for you because your depth of understanding will improve and others also will benefit from this work and most importantly read others blogs and code there are a lot of great blogs there is a lot of grey code that you can learn a lot from this is the strategy that I would recommend anybody who is learning machine learning and here I teach a very very applied a and machine learning course at apply day a course comm and all for all the topics that we discussed we have we have detailed contents on our website so if we go to her website here and if you just go to check out or free videos if you click on this link you will get this link now on this link you'll notice that we have some free videos please check out the free videos and give us feedback we would love to hear and we have designed a course with no prerequisites at all so if you if you scroll this you get all the contents here you get that there are hundreds of contents everything from what is up what is the equation of a line to deep learning techniques so if you just scroll down you'll get an exhaustive list of all the topics the detailed list of topics that you could benefit from if you're learning machine learning and we have about 140 hours of content explaining each of these techniques intuitively wherein we try to balance between both theory and practice we try to give a very deep intuitive explanation with lots of diagrams and we also do 10 plus case studies that we solve end to end so that you understand how to solve a real-world problem into N and we work with all of our students to build a portfolio with at least five case studies because that's what they can showcase to move into careers in machine learning and AI and as always as you might understand from my experience of Amazon I have learned customers the importance of customer service at Amazon and we provide and we aspire to provide we actually provide a terrific customer service because for any online course if there is no customer service customers may get stuck or students might get stuck at a concept and not understand so we provide both phone service and an email service where we try to answer customer questions as soon as possible there are a bunch of free videos on our website at applied a course comm as we just showed you so just go to applied a course come here and click on the click on the check out the free videos you'll get all these free videos and we would love to hear your feedback on what you think about our course and all the very best I hope some of the some of the solutions are given some of the material that I've given here or some of the some of the references that I've given here would be helpful in your in your in your in your learning experience of learning machine learning in the air and most importantly remember learning AI is not at all hard if you try to if you try to break down the mathematics into into diagrams if you try to break down the mathematics into the the dense equations into poetry this is my fundamental philosophy that equations in mathematics are like poetry geometry is like art and if you are willing to put in the effort if you're willing to put in five to ten hours a week for three to six months I think effort is the only prerequisite everything else can be learned in in life so all the very best in in your in your endeavors to learn and Excel in machine learning in hereso a question that is often asked is how to learn machine learning and AI and how much time does it take and how to actually what is the strategy to learn and my answer here is based on my experience as an instructor at apply day a course comm where we teach machine learning and applied and artificial religions from a very applied standpoint it's also based on my experience as a senior much learning scientist at Amazon my experience at Yahoo labs and startup that I founded where an I'm entered I meant a lot of youngsters in in moving to machine learning careers and in learning advanced techniques in machine learning and AI and the basic strategy to learn machine learning is this you need to learn both theory and practical aspects you can't just say I will learn only theory all I've known or I'll only learn some api's and learn how to program things if you just learn some aps and learn the practical aspects and if you don't know theory you'll always be lacking it's always important to get a very very good intuition on each concept not just the theory not just equations but also to understand each concept intuitively if needed through geometry I always say this I always say that equations in mathematics are like poetry and geometric intuition or geometry is like art right so if you can understand the concept both using equations which is like poetry and geometry which is like art your your intuitive grasp of things because we humans are very very visual creatures and we often gain better intuition when we try to visualize things using geometry the most important of all whenever you learn a technique in any area whether it's probability whether it's statistics whether it's machine learning anything trying to practice practice practice try to apply the technique lots of places to understand where it works where it doesn't work and for every technique you understand try to understand what are the boundary cases what happens in the boundary cases what are the failure cases where does a given technique fail and from my own experience of teaching lots of students machine learning in AI believe machine learning is not very hard lot of people panic when they see mathematics but remember mathematics is beautiful we can make it extremely simple using geometry if you use because V has visual creatures if you try to understand lot of people panic when they see big equations but if you try to read equations in English it becomes like poetry if you try to understand geometry it's like art you can appreciate mathematics better visually and also three equations right I believe I believe that learning machine learning and AI is not at all hard if you are willing to put in the effort the effort is something that we cannot run away from I believe if you can put in five to ten hours of effort of e you can learn machine learning in three to six months time of course there will be some fast learners who can do it faster there will be some slower learners who will take more time but on average we have seen that if you spend ten hours a week you can learn it in three to four months similarly if you put in five hours a week you can learn it in six to seven months and this is based on experience of some of the students that that we have taught over the last few months now this is this is a list of topics and I will give you some references which will help you so the first topic of course is you need to know a programming language preferably Python or R so Python is a very general purpose language R is a very statistical language my personal preference has always been with Python because it's much more easy to write code for those of us who know some programming language like C C++ Python should be very very easy to pick up there are these two very nice books there's a book called learning Python by Mark loads which gives you the overview of the language itself everything from what is a class what is a variable what is the dictionary etcetera - there is a book there's a very specific book by aurélie called Python for data analysis web by wes mckinney this does mostly data analysis term in this you learn specialized libraries like numpy site by matplotlib etcetera etc which are very very useful for machine learning and data science and in general artificial regions the second very important topic is probability and statistics some of you may panic as soon as you see propagate statistics but it need not be hard there is a very nice book called thing stats by ellen downey i believe this is also an overly textbook where where where the wherein the author takes a real-world example and tries to introduce all the concepts in probability and statistics using real-world data by you by asking very simple questions about data it's a very very interesting book especially for beginners and there is an online course called statistics and probability by Khan Academy Khan Academy is one of my favorite resources on the Internet because it because the explanation of concepts on Khan Academy is terrific some of you may realize that what I'm using the interface I'm using is also somewhat like Khan Academy not exactly like Khan Academy and I've learned a lot from Khan Academy over the years the third most important mathematical prerequisite is linear algebra for linear algebra you can easily connect linear algebra to geometry so understanding linear algebra concepts like what is the equation of a plane like what is the distance of a point from a plane we can understand all of that using geometry and the two great sources there is an online course on Khan Academy called linear algebra which is very interesting also there is a very nice book and there's also an online course called introduction to linear algebra by Gilbert Strang Gilbert Strang is a professor at MIT and this book is very very theoretical but it gives you much more mathematical rigor and depth right so but I don't know how much a practical aspect you might learn from this from this book but the linear algebra by Khan Academy is a very very but practical and theoretical it gives you a very well-rounded understanding of the concept now the fourth important thing is calculus and numerical optimization for this for calculus itself you need to know basics of calculus you don't have to know how to solve partial differential equations and advanced concepts basics of calculus what is differentiation what is partial differentiation these simple concepts are good enough and there is a course called multivariable calculus at Khan Academy which is which is a very good course and if you want if you prefer a book or a or a more rigorous mathematical course there is a online course called convex optimization there's also book by the same name by professor at Stanford called Stephen Boyd the first few chapters in this gives you a very very intuitive understanding on how optimization works and what Wow and there is both the book they're also very nice slides for this when I learned optimization myself in my grad school I learned a lot from Stephan Boyd's book and now after we have learned all these basic concepts we move on to machine learning we have a bunch of techniques called classification and regression techniques we also have techniques called clustering techniques and matrix factorization techniques and recommender systems for all of these sources for all of these topics there is a very popular online course called machine learning by andrew inc at Coursera it's a very popular course and there is also a very popular book called pattern recognition and machine learning by Christopher Bishop this book is very rigorous in its mathematically very deep well the online course is very introductory course which with some mathematical detail not too much and some applicative detail again not too much of applicative detail it tries to balance out both of them to some extent and this this book the pattern recognition and machine learning book by Christopher Bishop is a very good introductory textbook which is used in lot of colleges and universities so once you learn classification regression technique clustering techniques matrix factorization and recommendation system techniques you also have to learn some dimension reduction techniques there are a couple of dimensional reduction techniques which are very important to understand data then comes one of the hottest areas in machine learning today called neural networks and deep learning and for this there is a very popular course again online called deep learning by and ruing at Coursera which is very popular I also loved this book this book is one of my favorite books that I've read in the recent past it's called deep learning by an good fellow a very good book I've read this book ever since it was in preprint I really enjoyed reading this book it's it's a it's a it's a heavy book but it's worth reading it if you have bandwidth and for all of these techniques in addition to all these resources the most important thing to learn for each technique learn the mathematics learn the geometric intuition if there is a geometric intuition understand what are the assumptions that each of these techniques is making for it to work properly understand try to draw geometrically if you can or mathematically what is the best and worst case for each of these algorithms try to understand interpretability of each model interpretability for some techniques could be feature importance try to see how you can interpret a model don't ever look at models as black boxes because that way you'll never gain a deeper understanding understand how outliers or noisy points can affect your model understand how bias data sets or skew data sets are are handled in each of these modeling techniques also if you have time please code up each of these algorithms on your own that gives you a very very deep understanding of how each algorithm works for each algorithm there will be minor to major variations learning at least some of the variations will give you a better understanding on how on what what modifications can be made for a given algorithm so similarly for every technique it's extremely important to understand the limitations and how to work it on those limitations lot of people know a bunch of techniques but they don't know where not to apply the technique that's extremely important if you want to really apply machine learning to real-world problems the most important part is problem solving you might learn all the techniques or not but if you don't apply all of those techniques to real-world problems it's useless so tackle calm is a terrific website which which have which has competitions for machine learning scientists and research engineers and data scientists it's a great source where you can get a lot of datasets real world datasets I suggest you pick a bunch of problems from with a cool calm and solve these problems end to end solve it from how to pre-process data to trading multiple models and understanding this is extremely important understand why a technique or a model is working or not working because sometimes people just become coding monkeys we just try one model of the other model after a third model that's why you're not getting the intuition of why a model is working or not working or one why one model is better than other model try to answer those questions either using data visualization or using mathematics and for every problem that you solve please document and code up things and put it on a github or write a blog about it I strongly recommend anybody who is looking to move into a career in machine learning to build a port for you build a portfolio just like artists and photographers and models build a portfolio of at least five plus case studies and put your code explain your analysis draw plots explain why your technique is working why your technique is not working do a thorough analysis of each of these problems and write it it's it's useful for you because your depth of understanding will improve and others also will benefit from this work and most importantly read others blogs and code there are a lot of great blogs there is a lot of grey code that you can learn a lot from this is the strategy that I would recommend anybody who is learning machine learning and here I teach a very very applied a and machine learning course at apply day a course comm and all for all the topics that we discussed we have we have detailed contents on our website so if we go to her website here and if you just go to check out or free videos if you click on this link you will get this link now on this link you'll notice that we have some free videos please check out the free videos and give us feedback we would love to hear and we have designed a course with no prerequisites at all so if you if you scroll this you get all the contents here you get that there are hundreds of contents everything from what is up what is the equation of a line to deep learning techniques so if you just scroll down you'll get an exhaustive list of all the topics the detailed list of topics that you could benefit from if you're learning machine learning and we have about 140 hours of content explaining each of these techniques intuitively wherein we try to balance between both theory and practice we try to give a very deep intuitive explanation with lots of diagrams and we also do 10 plus case studies that we solve end to end so that you understand how to solve a real-world problem into N and we work with all of our students to build a portfolio with at least five case studies because that's what they can showcase to move into careers in machine learning and AI and as always as you might understand from my experience of Amazon I have learned customers the importance of customer service at Amazon and we provide and we aspire to provide we actually provide a terrific customer service because for any online course if there is no customer service customers may get stuck or students might get stuck at a concept and not understand so we provide both phone service and an email service where we try to answer customer questions as soon as possible there are a bunch of free videos on our website at applied a course comm as we just showed you so just go to applied a course come here and click on the click on the check out the free videos you'll get all these free videos and we would love to hear your feedback on what you think about our course and all the very best I hope some of the some of the solutions are given some of the material that I've given here or some of the some of the references that I've given here would be helpful in your in your in your in your learning experience of learning machine learning in the air and most importantly remember learning AI is not at all hard if you try to if you try to break down the mathematics into into diagrams if you try to break down the mathematics into the the dense equations into poetry this is my fundamental philosophy that equations in mathematics are like poetry geometry is like art and if you are willing to put in the effort if you're willing to put in five to ten hours a week for three to six months I think effort is the only prerequisite everything else can be learned in in life so all the very best in in your in your endeavors to learn and Excel in machine learning in here\n"