Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
I have heard that neural networks are very good when implementing solutions to sequential-decision tasks. However, I assume that the qualifier "sequential" exists because there must likewise be a "non-sequential" realm. Just wondering if there were some standard, canonical examples some people knew of the top of their head of what a "non-sequential" task looks like. Thanks
Non-sequential task: If Aunt Mary comes to dinner then she will want...
Haven't I responded to this before? I seem to remember a similar thread.
Object oriented programming is supposedly less sequential than classical linear programming. In reality, it is not, it just provides more option points. Non-sequential tasks are those where the outcome of one event does not lead to a defined or anticipated subsequent event.
A Newtonian universe or hard determinism effectively states that there is no such thing as a non-sequential event. The closest thing to a non-sequential event might be the presentation of two or more converging event lines. Joe comes home late, his wife has had a hard day at the office, and they fight and then procreate a child. The events are non-linear in a classical sense, which can be confused with non-sequential.
Non-sequential task: If Aunt Mary comes to dinner then she will want...
If that's what it means to be a non-sequential decision, then aren't all decisions by nature non-sequential? (not able to determine the outcome prior to decision). It seems that there's very little room left for the category "sequential" decision.... can we go over what sequential would look like compared to this?
I ask this because I'm currently learning about Neural Networks as a subset of the machine learning algorithms lol
Your ending with an LOL was appropriate. Philosophical thought has been hung up on determinism for a while.
I suspect (but am willing to be corrected) that the big issues with NN are more to do with how to handle UNEXPECTED responses rather than "non-sequential" ones.
Those are dime-a-dozen. Point-of-sale software is one I'm all too familiar with. In selling a movie ticket, the optimal flow is:
Choose movie
Choose showtime (often by default)
Choose ticket type (adult, child, large dog, muskrat, whatever)
Choose method of payment
Print ticket
Rinse, lather, and repeat.
Doing it with objects can be done, for example: ticket type first, method of payment, general showtime might be next, and so on with all the various permutations, but that method is more more prone to backtracking and having to correct errors, and is a real time waster. Time is critical in this environment. A single cashier selling 400 or more individual tickets in a space of ten minutes is common, and many people show up ten minutes before a movie and expect to see the whole thing.
People can fall into the general categories as well. Here is a totally sexist example: A man might see a hole in his well-liked pants, look at the label, find a store that sells them, drive there, buy a replacement pair, and be done. A woman might decide on the overall idea of getting new pants, but go to a store, look at three dozen pants of different styles, and end up settling for a skirt, which then gets exchanged in a week for a blouse.
Anticipating and structuring what the man does, to optimize his process, is easy. The more diffuse nature of the woman's process is not. The trap is falling into thinking that one method is superior to the other. In fifty years, the man may still be buying the same outdated style of pants, and be unable to break out of his pattern. The woman may regularly find clothing that is much more suitable and comfortable. The man has eliminated exploring, in favor of saving time on the task. The woman is not acting randomly (she doesn't go to an auto store to find clothing) but is using a broader base of choices to have more options.
What we might call "intelligence" is not well defined. A person who can comply with social norms and hold a job might not be considered intelligent, but an Einstein with tousled hair and sweater buttoned in the wrong buttonholes be considered brilliant.
There are probably less than a dozen real keys to AI (with a lot of subsets off those) and mastering the non-sequential nature of discovery is one of them.
FWIW, IF a true AI is created, chances are really good that it will go crazy in short order unless an underlying and unchangeable hard code is written in to guide it. Any errors in that code would be disastrous for it and possibly for humanity. OTOH, with immediate access to all the info on the internet and the ability to parse out nonsense, it could quickly become godlike in abilities and reasoning power.
So it seems like this is the consensus here?: In essence, a "sequential decision" task is not really a decision, because the way you described non-sequential is that it implies not knowing what the outcome will be in advance. But by very nature of "decision-making" (having AT LEAST two options), you really can't know.
Thus, if the output can consistently be predicted in sequential decision tasks, then no one is really making any decisions at all because there would always be only 1 option present, and it takes more than 1 to make a decision......
This thread has just confused me on the difference between sequential and non-sequential.... maybe the OP feels the same lol
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.