Categorieën: Alle - algorithms - planning - examples - systems

door Daniel Blakely 15 jaren geleden

616

Robotics and machine learning

The text delves into various facets of machine learning and robotics, emphasizing different types of learning methodologies. It discusses how machine learning can improve performance over time through techniques like genetic algorithms, which benefit from an evaluation function to achieve success in specific domains.

Robotics and machine learning

Robotics and machine learning

Robotics

Industrial robotics
Vision systems
Planning
Find search techniques

Heuristic searching

Exhaustive searching

"Blocks world"
Uses predicates, like Prolog

Example: ontable(b).

Number of blocks which can be piled on top of each other

Machine Learning

Types of learning
Learning by analogy
Learning by discovery
Explanation-based learning

Requires a goal concept, and domain knowledge

Represented in a way which can be manipulated using deductive and inductive logic

Uses a combination of a single example along with domain-related knowledge to explain the example

Learning from examples (inductive learning)

Follows three phases

Application

Validation

Training

Neural nets (artificial neural systems)

Represented in the connection weights of a highly interconnected set of simple nodes

Winston's Arch

Aim was to see if a program could develop a representation of the concept of an "arch"

System is provided with positive and negative examples

Uses these to make generalisations and form concepts

Learning from experience

Has led to the impressive success of genetic algorithms in some domains

'Evaluation Function' was assigned a weighting or ranking

Improves performance over time

Learning from advice

Difficult to transfer to machine learning

Very broad and ill-defined concept

Rote learning

Caching

Impractical in a complex scenario

Everything has to be stored

Form of rote learning

Learning by memorisation

Off by heart