the subtle humor in between makes the course so much more engaging.. One of the greatest instructors i've ever seen..i can't believe i completed a 1.5 hr video in one go.. kudos to Cassie, admirable job..
You know its great when after reading all the other stuff on ML and endless online material out there that this is the one that gives you that feeling of the "the penny has dropped" and "aha" moment in your understanding - great course. If you are new to ML or even have started on your journey, starting here will give you a great foundation to build upon your knowledge. It's even a great place to revisit concepts to refresh yourself. Thanks Cassie for sharing this with the world and help others in their journey.
Wow I'm joining Google next month as a intern & have always wanted to pursue ML but was afraid that I might not be competent. This video really encouraged me. Quote "if you're too afraid to start you tend to not do well in applied machine learning". Ur presentation was really great and spot on. Thanks for the dedication in making it easier for us.
This is marvellous! Cassie, I'm in the middle, but already remember more about basics than after 5 previous ML courses I've already did. And it's funny. And I love how you talk about this stuff. First time in years I want to watch more about ML and I don't feel bored after 15 minutes. I will definitely watch all your lectures and finally apply all the knowledge that I'm gaining here. Because what I am missing is the flavour and a concept. I now the math for ML, but I lack the feeling, I think the most important part. Bravo Cassie! I'm so happy these lectures are available.
If every course or subject was taught like that. I'd be an A+ student. Excellent, fun, not boring at all, a joy to watch and learn from course. Thank you so much and please keep doing more of these 😁👌🙏
I am absolutely impressed by the way you express all the fundamentals, it's just simple and powerful. Your analogies and the way you present them are beautiful. Thank you so much for your time and effort. I'll show this to all of my friends who wants to have some ideas how AI and ML work.
as a statistician who has been rather skeptical of the entire premise of 'artificial intelligence' i must say this lecture series got me interested in understanding a little more. i think i still have many many reservations which probably stem from familiarity with models which can serve both for prediction but also for inference. whereas with these models, it seems like we are trading away inference and even the ability to analyse the model behaviour in pursuit of better and better quality prediction. i think the work you are doing here to communicate that tradeoff is important. thanks for making this available for free.
This is wonderful. Had to pause the video because after the ML definition all the previous examples she gave were just the ideas to explain the best possible definition. Very smart and now I need to review everything but so far very entertained yes, many technical topics well explained but need to be digested. So far loving it 😍
The RU-vid logic will most likely recommend to all of my friends in India especially in the IT field. Most likely this video will hit 10,000 views in another 3 months . This will work because RU-vid recommendation logic will do it's job like a charm
Love the approach of taking a complex topic and doing this for dummies version. Moving on to the next lecture … thank you for making these available for the wider world
I am really attracted to and motivated by your way of presentation, teaching, and making people see what is taken as hard become easy. You are amazing, Thanks for this talk.
Really appreciate you making this public. A great presentation; really made me think differently about how we use data science in business. I've followed your posts on LinkedIn and just stumbled upon your RU-vid... I'm glad I did!
I can say this is the best course for ML till now. Most of the folks doesn’t know / understand / realise there is two ML types research and Applied. As a designer I am tinkering with fusing design thinking and ML for past 3 years and I am just getting there to work up a process recipe so that anyone can become a wonderful chef & it start with understanding the business problem and user problem. This course helped me understand how what I am doing is super important in the realm of the things today and tomorrow. Thanks Cassie 🙏🏻🤟🏻
Hey Cassie, Such a wonderful video. Thanks a ton again for making it available for public. I am watching it for third time :) coz i keep understanding more every time. One request if you could make it happen - there are irritating videos which keep popping in between, not sure if you can remove it. Thanks Again, Appreciate the super good content and your calm way of explaining concepts.
I really relate to the issue of people lacking the ability to apply concepts in the real world. After my MMathStat I felt like I learnt how to pass exams with little idea how to do these things in the real world.
Incredible! You speak really well and use wonderful illustrations. Thanks for sharing (I found you by going through Google’s data analytics certification on corsera, so glad I dag through that one page about linkedin even though I didn’t need it lol)
Especially the kitchen one towards the end and how you differentiated between research and applied AI/ML. Good microwave engineers are probably horrible cook lol
One question: how AI will decide if the data is based on two opposite examples, such as the example of the Ethics of Kant (Deontological Ethics: shall say the truth always) and the Ethics of Benjamin Constant (Consequentialist Ethics: the truth only according to the consequence)?
Hi Cassie, @1:17:52 Given that at any point in time there are 10 empty spaces on average, X_bar = 990. Lets say there is 1000 trials of the technology and a requirement of 95% accuracy before the green light: With the strategy of full capacity, 1000 spaces occupied and the accuracy requirement of 95% given that X_bar = 990; there is only two ways that the strategy would be a winning strategy: i.e. what combination of the sample space(1000) will lead to X_bar = 990? The strategy needs to be right 95% of the time Therefore 1000*950 = 950,000 There is 50 sample spaces left in the trial Lets say there's a brilliant optimization technique employed and it outputs: 800 * 50 or (900*40) + (400*10) X_bar = ((1000*950) + (800*50))/1000 = 990 or X_bar = ((1000*950) + (900*40) + (400*10))/1000 = 990 So out of all the possibilities that will lead to X_bar = 990 there is only two ways given the strategy constraint. The problem therefore is the likelihood of the strategy given X_bar, it's too extreme! On learning, I think your student was right! Here's why: Children achieve mastery of their local language from an early age without ever sitting an exam. They're able to apply their learning directly in social interaction without testing because this is simply the structure of their social interaction. Would you agree to the possibilities of better alternatives?
I’m curious, how can we ‘improve’ our decision intelligence? You mentioned how that reliable workers (AI or ML) can scale up stupidity as well as intelligence, I really think that if I continue diving in this field I’m going to see exasperated people whom I thought I was going to impress with my answers lol. How do u suggest we improve at being better at decision intelligence?
I'm really very confused. When you created your wine bottle chooser recipe, you plotted out your real life data of preferences after an undoubtedly memorial evening. Then you drew a line to separate "yes" from "no". So whatever relationship between your wine critic opinions and wine age that this line describes, there is your decision maker. I don't see where there is a "black box" of algorithms. You"decided where to put that thing" based on your visual inspection of your plot. Just don't get it!
@datascientific Hey Cassie, thanks for this class! In the parking lot scenario, would would the better question be to ask our system of not “if the space is empty or full”?
@@kozyrkov First ever given, and first course of yours I didn't take live. I felt pretty cheap tbh, it's a great course, I've already applied it and I really appreciate your focus on practicality and keeping it engaging. Thanks for doing these!
She is a very good lecturer (brilliant!) but she failed miserably when she showed utter ignorance and disrespect for some very deep and groundbreaking work done between the 50s and 70s (‘Classical AI’) where some basic understanding of how our own brains work, how we understand the world around us and how we solve problems. Today’s “AI” has no “understanding” of science, physics or biology. All it does is pattern matching between vast amounts of data. And yes-it’s getting intriguing results - but also fails again and again when there’s a limited amount on information on a specific problem. I was expecting a more balanced lecture that would at least do some justice to geniuses like Minsky, Lenat, Newell, Simon, McCarthy, Shannon, Samuel and many others who surely wouldn’t look at today’s calculator and call it ‘intelligent’. This is plain demagogy. Makes me sad that an otherwise very bright scholar would be so uneducated and mislead the audience to think that the new ‘AI’ is more than what it is.