Intelligence is the ability to think rationally and to control action. Humans have the intelligence to think and use common sense to make decisions. Artificial intelligence is a field of research to build intelligent agents, so that in the future we can build ARTIFICIAL intelligence that can think like humans and act rationally. The Turing test was proposed by Alan Turing (1950) to provide a satisfactory definition of intelligent operation. A robot can pass the Turing test if it has:
1. Communicate with people by understanding and writing natural language;
2. Knowledge representation (knowing how to present knowledge to users);
3. Knowledge reasoning (knowing how to infer answers from stored knowledge to answer humans);
4. Machine learning to infer patterns and adapt to new environments.
In short, AI is about studying the rules and algorithms that help build intelligent machines, and the set of problems that AI solves is NP-complete.
Artificial intelligence is a broad field of research involving five important disciplines:
1. Expert system;
2. Neural network;
3. Fuzzy system;
4. Robot;
5. Natural language processing.
Machine Learning (ML)
Machine learning is a subset of artificial intelligence that uses algorithms to learn from data to get data that humans need. Learning can turn people into geniuses and adapt them to new environments. Similarly, machines’ ability to learn makes them strong enough to adapt to new environments. The goal of any machine learning algorithm is to maximize its goal through the learning process so that it can process invisible data.
Two key learning methods (algorithms) to implement machine learning are:
1. Supervised learning: External designers or tagging data facilitates machine learning.
2. Unsupervised learning: Machine learning without any tag data or external designers.
The goal of artificial intelligence is to make machines as smart as humans.
Expert system
Expert system is a system that relies on knowledge base to solve problems. Knowledge bases can be represented in different forms, such as rules, semantic networks, and decision trees. Expert system consists of knowledge base and inference engine to infer or reason knowledge from stored knowledge base. Expert systems are used where human experts are needed to solve specific problems.
The knowledge base
Rule-based expert systems capture expert knowledge in a specific domain in the form of rules. These rules form a knowledge base, which is then evaluated by an inference engine to solve a particular problem. Example rules:
If the sky is clear and the sun is shining,
A raincoat is not needed, then.
advantages
- Because rules are expressed in natural language, it is easy to capture an understanding of the knowledge base.
disadvantages
- Experts have different opinions on the same topic, which makes domain knowledge difficult to master.
- Maintaining and updating rules is a long process.
And there are different types of expert system in different fields, such as rule-based expert system, fuzzy expert system and frame-based expert system.
reasoning
Reasoning in expert systems is done by linking forward or backward. Forward linking is a data-driven reasoning technique that starts with knowing the data and follows the rules. Backlinking is goal-driven reasoning that starts with a goal and works backwards to find data that supports that goal.
The neural network
Artificial neural networks (ANN) are inspired by the human nervous system. The system works in exactly the same way that the human brain stores and processes knowledge. A neural network, much like the human brain, consists of a group of neurons, or nodes, that are highly connected to each other. Information is stored, processed, and analyzed in the neurons of the network. Each node or neuron can activate other neurons in the network, and the links or connections between neurons are called weights. A network can contain n neurons or nodes, which can make the network very complex. A simple neural network consists of an input and output layer.
Here are the different types of neural networks:
- Feedforward neural network;
- Convolutional Neural Network (CNN);
- Recursive neural network;
- Long short Term Memory Network (LSTM).
Artificial neural networks can learn by adjusting their weights. It is this power of neural networks that makes them suitable for machine learning. Different types of learning algorithms can be used in neural networks, among which the most prominent is the back propagation algorithm.
The original link