Re: Negative Reinforcement
Great, and how is that mechanism different from the adversary's regarding your positive reinforcement - aka reward - as his punishment? Same problem, with a changed sign.
35 posts • joined 13 Jul 2015
What is the big deal [of computers beating a human at Go]? Is this the purpose of Go [to test humans' brain power versus computers' ability to create all - or most of - the combinations]?
Next, Facebook will be simulated within facebook to learn who spends most time on Facebook. Then within that simulation there will be another one and so one --- all while the oil price will go up b/c so much energy goes to play Go, and simulate [well, try to predict the percentage of] users who will volunteer their nude pics to Facebook. Sorry, but I cannot do that - says Dave, for a change!
give them a chance, man - don't get mad, use this:
Shapiro, Stuart C. (1992). Artificial Intelligence In Stuart C. Shapiro (Ed.), Encyclopedia of Artificial Intelligence (Second Edition, pp. 54–57). New York: John Wiley. (Section 4 is on "AI-Complete Tasks".)
Just kidding, it won't be another winter! Too expensive and disappointing for all the Singularity adepts... They can't be wrong! (or rather one cannot prove them wrong) But I bet the notion of AI-completeness will be re-phrased, re-thought, re-worked, re-defined, re-invented, and retracted.
This validated machine learning prediction tool told me - on a test data - that I will win the lottery tomorrow; on the training data it said I had already won yesterday. Hence, today's prediction is: I will keep my job.
And talking about NLP: do I say that I will keep my job or the machine predicts that I will keep my job? What is it then with the lottery?!
Are you telling us that if we build 50 beellions artificial neurons they will not be able to write poetry or compose a symphony? You've got to be kidding! Computers (deep learning to be fair) are already producing musing - you may call it cacophony and ouch! I will have to agree!
30 years ago it would be called testing - and a good tester would know where the system breaks. Later on it would be called V&V (referring to the SW used to analyze data and make decisions).
Driving autonomously is an AI-complete problem and only after it will be accepted as [nearly] impossible, a [close to bullet-proof] solution will "emerge", and will called the rail road. And it will be that, too.
None- all public info- it just needs a few 13 year old to code in Python ... b/c nowadays we don't really care about optimization - what is a microsecond [reaction] time delay? This is where we are with scientific computing these days ... again, what is the big deal !? Buy the sensor, the code is done in no time.. What is not going to be done in no time is the realization that self-driving is an AI-complete problem, and as much as I agree we are getting increasingly quality, we are not going to get it perfect - in some ways it will be better than human driving, in some ways worse. Only when self-driving cars will have the model of the railroads we can talk about 100% (well...) reliability .
And what is the definition of bad statistics? Or, rather, what is the definition of statistics? You only say what its purpose is (are, in fact, as you list two). We can then understand what is the difference between statistics and bad statistics, and where the study went wrong. In fact, it is nothing wrong with the study - it obeys statistics. We just forget (and forgive) the fact that s a medicine works in 99.999% of the patients, but it does not work on you, you will 100000% die (if we talk about an antibiotic to treat your flesh eating bacteria).
Let us try rather understand what could be the conclusions we draw from stats used to interpret weather data and those used to interpret data that will determine the percentage of people who will get a certain disease. If the weather prediction is wrong, the consequences could be: take an umbrella on a sunny day (mild annoyance), all the way to putting you in the middle of the twister. Ouch! If the data on disease say that only 0.02% of the population will get that, then you are at a low risk, indeed, of getting the disease. But if you got it, there is a very high likelihood that there will be no studies and dedicated medicine for it. Ouch again! So is this the fault of statistics? I cannot blame statistics, because there, on the first page of the book I have, it says clearly that winning the lottery is not guaranteed. So, use it with caution and with a clear understanding of what the interpretation of the data may be. I'd say, if we engage common sense, we'll be fine.
I would have long played the lottery or the stack market, won, and went back to work for free without having to deal with managers. No, don't tell me that I can just work for myself :) Leonardo Da Vinci was (still is) unique.
The mighty pdf (probability distribution function) has us for breakfast everytime we "predict" without data and without knowing how the data looks like.
We have not figured out how the brain works - or was I missing that big news while watching The Stepford Wives?
For those few, apparently, who skipped the biology class: the nose smells what gets *inside* the nose (yes, a surgery mask will keep you happy during the flight). So how would then the SW get those molecules inside *its* "nose" first then inside ours - I guess will have to call it smellware.
Before we move down to the mouth and I read in Science about the SW correlation and correctness of smell and taste: the taste buds do for the mouth/taste (and brain) what the cilia do for the nose/smell.
Anyone wants to bet that we will read about TW (tasteware) in the next 6 months?
Before the term "machine learning" was coined, it was "Statistical learning"; machine learning became prominent in the "learning systems" component of AI. Most machine learning techniques are - speaking in numerical analysis/methods terms - interpolations and extrapolations with a rarely known, yet too often imposed/assumed/accepted/shoved-down-the-throat pdf (probability distiribution function) or "the mighty pdf".
The author says" "Wherever you set the bar, I guarantee that you will find that the system you are calling AI is heavily dependent on machine learning, which only works if we have data mining, which relies heavily on statistics, which is fundamentally founded on maths."
Here is a counterexample from the times when the data was not so aplenty. Medical diagnosis systems of the 1970's or 1980's were rule-based, no machine learning - and they were AI, which began to have an impact as soon as computers were used according to the "definition" from the Association for the Advancement of Artifical Intelligence http://www.aaai.org/
There is no need to attempt to define AI so many times over. It is hard to give a definition of AI that would be accepted by everyone - so AAAI provides the standard - agree or not, let's use that one. Yes, there are people who think "AI is a small corner of machine learning" - but only after mastering, for example, the book by Hastie. Tibshirani, and Friedman "The Elements of Statistical Learning"
http://statweb.stanford.edu/~tibs/ElemStatLearn/printings/ESLII_print10.pdf should we speak about where machine learnig fits in regard to AI, math, life, Earth, Galaxy, etc
Fukoku Mutual Life Insurance has an AI system - because they likely use a statistical learnign/machine learning to exploit the data. If rules are provided for any possible scenario, it is TurboTax for insurance - and it is AI (TurboTax is an expert system). Imagine life w/o TurboTax, with everyone's tax done by a knwoledgeable person. Now we talk millions of jobs lost to AI! Since before the term AI was omnipresent.
1. Is this "work" just a recycling of Charleston Heston's The Omega Man?
2. Statistics in physics are great, but taking that approach [an assumed probability distribution function] is not necessarily working in every aspect of life. If it would, we would play the stock market successfully, then work for free -- for example, for those universities where like modeling -- and we will use a non-Gaussian distribution to show that the answer is, indeed, 42 - whatever the question.
Suggestions for next big article title:
"Physics departments are shrinking; WHY???"
"I just think he's biased. Taking a human point of view -- but humans do tend to be full of themselves until they screw up."
I just think we should watch you: I bet you come straight form The Factory for the Absolute. You don't like us, the humans! We know you are a machine! We screw up…. again!
To Destroy All Monsters:
1) you don't find them - you pretend you did. They do not exist - please check this out
http://static1.1.sqspcdn.com/static/f/702523/9242013/1288741087497/200801-Dewar.pdf?token=q4qOQEN1oIuukznezyyt29ndVSc%3D it is a problem long recognized and even longer ignored
2) nothing: you cannot pay what you do not have
3) you don't need to- but if the number is a must, you can hire anyone who can spell the word "cloud" and go back to 1)
Biting the hand that feeds IT © 1998–2022