An introduction to machine learning (a form of artificial intelligence concerned with getting computers to learn from data), and a discussion of some of the mathematics underlying machine learning algorithms.
Part 1 of 4:
Part 2 of 4:
Part 3 of 4:
Part 4 of 4:
Here’s a very nice introduction to machine learning by Ian Murray.
What area of mathematics would study the concepts in these videos? (Not the idea of machine learning which might be from Computer Science) but the idea of error, minimizing functions, ect.. Is it statistics or is there a more precise sub-feild within stats that deals with this?
The idea of error comes up a lot in statistics, though it is often treated a bit differently there than in machine learning (for instance, in statistics you’ll frequently see the assumption that prediction errors made by a model have a certain distribution are are independent from each other, something that I haven’t seen a lot of in machine learning). The study of finding the minimal value of functions or finding the optimal parameters to minimize a function can be found in the field of optimization theory.
You are very generous when it comes to sharing your wealth of knowlege. I really enjoy both your blogs. Thank you !
Query: Before you decide to utilize a machine learning function, you have to decide the parameters of its primary objective, you can’t just create a brain and then set it loose on the world, even humans have instinctual bias hardwired into their systems, we can choose to ignore or follow, but as babies, we lived by these doctrine, without them, we would have died by the flood of information but not knowing what to do with it…
so what is it you all are trying to do, create an intelligence or create a prediction machine that can react to unforeseen circumstances?
Where there is Artificial Intelligence, there is Artificial Ignorance, e.g. Boeing altitude detectors…
Hunan intelligence is multilayered from different parts of the brain, left, right and emotional. This (can) lead to a self- realization of what we do not know. (Except
to the “Know It All” personality.)
This ability to understand what one does not know is an important aspect of Wisdom, that AI does NOT have. Thus the Boeing crashes.
All AI is nothing more than output based on arrangements of “Zeros” & “Ones” in the logical patterns of Boolean algebra. It was originally used by an early Computer Co. to Select & Arrange the drawers of electrical components of main-frame computers for specific orders. (I believe I recall it was early Univac.) In that case the logic
of allowable arrangements was defined & limited by the computer design engineers but was complex and time consuming to complete manually.)
AI “learning” is limited to the same “rules based logic.” So, it will never be as intelligent
as a human, though brilliant in specific areas were external information is not relevant, e.g. Chess Games…
All the rest of this is false advertising. Keep in mind the Boeing problem.