# (Real) Machine Learning

Discussion in 'Intelligence & Machines' started by BenTheMan, Oct 3, 2012.

1. ### EmilValued Senior Member

Messages:
2,789
I think the translation is wrong and I do not understand exactly what is "Machine Learning"
It's about something like this? Finite-state machine

3. ### AlphaNumericFully ionizedRegistered Senior Member

Messages:
6,697
"Machine learning" is the area of mathematics/computer science/(engineering?) which attempts to construct algorithmic procedures whereby a computer might be able to process data such as video, sound, large databases and do 'human-like' things such as classify data, identify objects, infer about missing data, predict things, make associations between data structures, make decisions which take into account new data etc. All of these are important if you're looking to build a machine which is to replace a human in certain things.

On very simple levels you have things like identifying objects in pictures, identifying objects in videos, tracking their motion etc. It's a non-trivial task to identify objects in video and then track their motions correctly. For example, if the video is of a crowd then people will walk behind one another, blocking your view. How would a computer identify someone when they come back into view? It's a very basic task but a lot of the underlying mathematics has applications to more elaborate data types.

A FSM is a simple model of a computer or mechanical device. You list all possible states the machine could be in and all possible transitions between them. There's a few other things but basically you can model simple computers using FSM. It's very impractical once you scale to larger machines. No one models real computers that way.

Ben et al, I'll reply to you when I have the time this evening.

5. ### EmilValued Senior Member

Messages:
2,789
Yes, I understand now. I do not know how to do it exactly, only in principle.
During college,at the university, as stated by Billy T, we worked on analogue computers. It was a pleasant thing.
We do different dynamic systems models (simple or complex) with variable parameters, where time was variable and watching outputs as temperature or breaking strength or critical oscillations.
We were able to predict when a system is overheating, breaks or resonate.
Yes, FSM are automatic for some specific application with reduced mobility.
This was efficient when complex integrated circuits were expensive.
When these circuits were cheaper, we used motherboard microprocessor, which had a library such as input output protocols and also the monitor.
Programming was done in mnemonic (machine code). These automation were more mobile and can always rewrite the program.

7. ### pitrodautilityRegistered Member

Messages:
3
Hydraulic Cylindrical Grinding Machine

Core drilling is a process by which a cylindrical solid, similar to a hole saw, can be removed from a material. Core is the material which is left in the drill bit and this process has many applications. It can be carried out if one needs the core of the material or where rapid drilling needs to be conducted because removal of less material is to be performed.

Thanks.

8. ### leopoldValued Senior Member

Messages:
17,455
mnemonic and machine code are 2 different things.
machine code is the ones and zeros.
specific groups of these ones and zeros represent instructions and its these that are shortened to mnemonics.
you feed mnemonics to an assembler and IT generates the machine code.
it should be pointed out that hex is almost exclusively used with assembler instructions.
machine code programming went out the window with the MITS ALTAIR.