Machine Learning is the current and next natural step for research and development in artificial intelligence. Machine Learning has been described as applied statistics, advanced algorithms and other definitions, but essentially is a mix of all of these definitions. Artificial intelligence is about processing data to understand it and react intelligent to the information feeded in the mathematical model. In some way, Machine Learning is the natural evolution of all the research developed so far.
When software is written, usually the decision making is built directly into the program. For example, if a program is designed to recognize my face, The decision could be structured like this: Check for black hair > Check for brown eyes > Check for glasses and etc. In the other hand, with Machine Learning, a mathematical model is built to look and process a bunch of images over time to find a balance and generalize the information. Therefore, if another image is processed the model will respond: *This face is Victor, This face is Brian, This face is Paco, This face is John, This face is Carlos, and so on. The model was able to recognize these people based on the previous experience (the images, given to the model previously), hence is crucial the data used to "train" the Machine Learning model.
Currently Machine Learning techniques are been applied to every domain: Improve business decisions, identify diseases, distribute water in cities, coordinate traffic lights, search planets, etc. Applied to complex problems to manage uncertainty.
Search techniques tree diagram. (Some people might think differently and situate AI in a different branch)
GENERAL PROBLEMS IN ARTIFICIAL INTELLIGENCE
- intelligent models have limited resources
- Computation is local, but problems have global constraints and influences, how to deal with that?
- Logic is deductive but many problems are not
- The real world is dynamic, but the knowledge is limited
- Problem solving, reasoning and learning are complex problems, but explain and justify that problems are even more complex
Main AI characteristic problems
- Knowledge and data often arrives incrementally, so the scope changes.
- Problems exhibits recurring patterns, many of them hidden from human view.
- Problems have multiple levels of granularity.
- Many problems are computationally intractable.
- The world is dynamic but you face the problem from the static point of view, because your knowledge of the world is static or at least does not change fast as the problem you want to resolve.
- The world is open – knowledge. In contrast, our understanding is limited - knowledge.
INTELLIGENCE BEHAVIOR PROCESS
There are 3 form of process which are intrinsically connected to perform intelligent behavior. These are: Reasoning, Learning and Memory and they constitute a unified architecture. Each form of process feedback in both directions: We learn because, we could store the experience in our memory and that allow us to reason. The more we know, the more we can learn, and so on. When a joke is says, it always depend on a context and we store in the memory people reaction. When a similar context occurs again the brain recover from the memory the people reaction and the reasoning allow us to say the joke again or not. We learnt when is funny to say the joke.
There are many knowledge theories based on Artificial Intelligence that unify reasoning, learning and memory. The idea behind to put this 3 elements together is called deliberation, and deliberation is only one part of the overall architecture of Artificial Intelligent agents knowledge.
Reasoning: Understand natural language sentences. Generates natural language and make decisions.
Learning: Answer the right questions and store the answer somewhere, then connect and amend the wrong answers.
Memory: When you learn something, that knowledge that your learnt has to be stored somewhere (memory), If you are going to reason using knowledge then, that knowledge has to access from memory. The memory will store what we have learnt as well as provide access to knowledge for reasoning.
Intelligence behavior process diagram
Machine Learning diagram ecosystem emulating: Human-Thinking, Human-Acting, Acting-Optimally and Thinking-Optimally