Learning
Learning adds another function that the percepts must fulfill; The percepts must not only be used for acting, but also for improving the agent's ability to act in the future. Learning takes place as the agent observes its interactions with the world, and it's own decision-making process.
There are so many ways to categorise learning, and the fact that some categories overlap with others can make things a little confusing. To further confuse the issue, different authors use different terminology, and hence inconsistencies exists between different texts. What is to follow will attempt to clarify things as much as possible, hopefully not (but probably) in vain.
Learning Paradigms
Learning Algorithms
Haykin describes learning in the context of a neural network as follows...
Learning is a process by which the free parameters of a neural network are adapted through a process of stimulation by the environment in which the network is embedded. The type of learning is determined by the manner in which the parameter changes take place.
A prescribed set of well-defined rules for the solution of a learning problem is called a learning algorithm. Again, in the context of a neural network, a learning algorithm is defined by the way in which it makes adjustments to the synaptic weights of a neuron.
Abstract Learning Algorithms
- Error-Correction Learning?
- Memory-Based Learning?
- Hebbian Learning
- Competitive Learning?
- Boltzmann Learning?
Concrete Learning Algorithms
- Back-Propagation Algorithm (supervised learning)
- Genetic Algorithms (reinforcement evolutionary learning)
- TD Learning Algorithms (reinforcement evolutionary learning)
- AHD Algorithms
- Sarsa Algorithms (On-Policy)
- Q-Learning Algorithms (Off-Policy)
Learning Tasks?
- Pattern Association?
- Pattern Recognition?
- Function Approximation?
- Control?
- Filtering?
- Beamforming?
- Beamforming?
Learning Machines
- Decision Trees
- Neural Networks - Artificial neural networks are largely used in the realm of supervised learning, however they are also used in reinforcement learning and unsupervised learning.
References
Haykin, S. (1999). Neural Networks, A Comprehensive Foundation (2nd ed.). Upper Saddle River, New Jersey, Prentice Hall. Chapter 2
Related
Ancstors ☣ ..
Other ☣ Algorithm/Learning ☣ Agent/Learning ☣ Learning/Machine/Data
