Machine Learning Course - 13. Intelligent User Experiences
Ғылым және технология
A full university-level machine learning course - for free. New lectures every week.
Designed as a first course for engineers, program managers, and data professionals who want to learn: the details of important machine learning algorithms; professional model building; and design patterns for building practical machine learning systems.
This lecture covers: How to connect machine learning models to user experiences, how to balance the quality of the model with the forcefulness and frequency of the user interaction, and other techniques to use to get the most out of your models.
And you can view the entire course in this playlist: • Machine Learning Course
Learn more (and find slides for this lecture) at: www.livingmachinelearning.com/course.html
Learn about machine learning UX for your intelligent systems.
Пікірлер: 3
bRILLIANT POINT DR. AT 5:41, I have built a model for electric theft prediction for the electric utility, with accuracy of 94%, and published a paper too. , I was very happy till this lecture but now you have highlighted the super point (mistakes correlation with a number of consumers). I was not thinking in that way, and most of the papers I have reviewed have not mentioned this too, so technically DR, what should we do if this situation happens, we do not publish the model till 100% accuracy which is not possible or it depends on the condition to condition
@GeoffHulten
3 жыл бұрын
Great question - thanks! For publishing, it depends on your research contribution. You don't have to solve every problem to have a valuable research paper - you just have to contribute something to the field. No model will be 100% - the point of this lecture is that you can control the way you present your model's predictions to users so that the correct answers are amplified and the incorrect ones are mitigated. This thought process is a key component to being a successful machine learning architect.
@SaddamHussain-qt4cm
3 жыл бұрын
@@GeoffHulten Thanks for the response. From a research point, you are right, we can not fulfill all the contributions. But imagine, a research paper published in impact factor of 9 to 10, the author is claiming to achieve accuracy of 90% , deployed that developed ML-model for million users, how pathetic and annoying the final experience will be. But I guess, from traditional "hit and trial methods", ML- methods are way much better but we have to be very careful and do not get into the hype. Thanks for the great lecture, you are giving free of cost. So many pathetic courses on udemy are there. They do not go into details still they have thousands of students, I don't understand the gold you are giving free of cost, i have read the book "The Death of Expertise Book by Tom Nichols"......which clearly explains how people get fooled and pay less attention to the Experts like you. I wish you keep on uploading, your channel will grow to millions.