MAI - topics 5-9 Flashcards
Problem (T5)
Situation, question, or thing that causes difficulty, stress, or doubt.
Problem Solving Process (T5)
Understand the problem - identify and define
Plan – analyse, form a strategy
Do – resources, organise information
Check – moniter progress, evalutate results
Does not need to be in order
Knowledge Representation (T5)
Representation of information from real world for a computer to understand and solve
DIKW Model (T5)
Data Information Knowledge Wisdom (pyramid)
Data (T5)
Representation of facts, observations and occurrences – raw without processing
Information (T5)
Data converted into meaningful form
Knowledge (T5)
Facts, skills and understanding gained, through experience, which enhances ability to evaluate context, make decisions and take actions
Wisdom (T5)
Ability to act critically or practically in a situation – based on ethical judgement based on an individual’s belief system – application of knowledge centred around a certain criteria
Knowledge vs Wisdom (T5)
Knowledge - collection of information through learning, skills and studies, organised information
Wisdom - ability to judge and apply education to practical life
Machine Reasoning (T5)
Process of taking information that is known along with background information to make inferences regarding unknown information
Knowledge Processing System (T5)
Allows a strategy to make a decision in acquiring knowledge
Machines struggle to interpret all human information – must represent
Declarative Knowledge (T5)
Concepts, Facts
Structural Knowledge (T5)
Problem solving, relationship between concepts and objects
Procedural Knowledge (T5)
Knowing how to do something (rules)
Meta Knowledge (T5)
Knowledge about other types of knowledge (planning, learning)
Heuristic Knowledge (T5)
Expert knowledge in a field – solves problems based on experience of previous problems
KR Components (T5)
Perception – retrieves data
Learning – learns from captured data
Knowledge representation and reasoning –
humanlike intelligence
Planning and execution – depend on analysis from KRR
Intelligent Agents (T5)
Anything that can be taken as percieving an environment
Structure of an Intelligent Agent (T5)
Sensing – perceiving features of environment
Thinking – decide what action to take
Acting – doing things
Machine Learning vs Reasoning (T5)
ML - mathematics, predictions, analytical, training, works on data
Reason - KR, problem solving, expert, works on knowledge
Optimisation problems (T6)
Huge proportion of problems we ask computers to solve
Best solution from a number of alternatives
Heuristics
Combinatorial optimisation problems (T6)
Find the best combination of discrete values from some given set
Combinational Explosion (T6)
Occurs in numeric problems when the complexity rapidly increases, caused by the increasing the number of possible combinations of inputs
Curse of dimensionality
Curse of Dimensionality (T6)
Explosive nature of increasing data dimensions and its resulting exponential increase in computational efforts required for its processing
Observed in machine learning
Increase in dimensions add more information – improve quality of data but practically increases noise as well
Symbolic AI (T6)
- Must provide all the characteristics in advance
- Able to replicate high level human capabilities
using recent advances on formal systems and computational capacity - Decision tree
Connectionist AI (T6)
Tries to produce intelligent behavior replicating neural structures: small computational units playing the role of “neurons”
Machine Learning (T6)
Field of AI that applies statistical methods to enable computer systems to learn from the data towards an end goal
Uses various types of automated algorithms that learn to model functions and predict future actions from data
Deep Learning (T6)
Neural networks with several hidden layers
- training (calibration) and validation (assess)
Uses neural networks that pass data through many processing layers to interpret data features and relationships - forecasting
ML vs DL (T6)
ML - thousands of data points, neumerical value, predict future actions, directed by data analyists
DL - millions of data points, numerical to free-form elements, interpretation, largely sefl directed
DL Feedforward Networks (T6)
Multi-layered networks, where inputs are weighted and some function transforms the weighted total giving the output, which is the input of the next layer
- use errors to learn
Deep Learning Applications (T6)
- Use of neural network for programming
- Classical dynamic programming – based on optimisation
- Reinforcement learning – constructing networks capable of optimal behaviour on a given artificial environment
Cognitive Computing (T7)
Try to mimic a human brain by analysing text/speech/sound/images and objects to give a desired outcome (AI-based language translation software)
Speech Recognition (T7)
Cognitive service that aims to replicate a human action - processes sproken word and processes it into readable text
Speech Recognition Applications (T7)
Transcription, voice-based search engines, produces data for analysis
Computer (T8)
Device that can be instructed to carry out sequences of arithmetic or logical operations automatically or computer programming
Robot (T8)
Machine which may or may not require intelligence, specific tasks, physical form
AI (T8)
Ability for a computer or robot to perform tasks associated with intelligent beings
Robot Types (T8)
Autonomous/semi-autonomous
Robotics Applications (T8)
Assembly, customer service, packaging, open source robotics (AI)
