AI, Machine Learning and Data Science
AI, Machine Learning and Data Science
• AI vs Data Science
• Differences
– Post Analysis
Data scientists typically interpret and report their findings,
whereas machine learning engineers focus on the model's
deployment, monitoring, and maintaining in a production
environment.
Data Science and Machine Learning
Artificial Intelligence and Data
Science
Artificial Intelligence
– Decision Making
• Problem Solving
– Maze game
– Tic tac toe
(No data no modelling)
(Need efficient algo. A*, DFS/BFS)
Knowledge Representation &
Reasoning
• Knowledge Representation & Reasoning
– Eight Queen’s problem
– complex game rules
(Write rules and apply reasoning)
(No data no modelling)
(Need propositional and first order logic.)
Decision Making
• Decision Making
Expert Systems (when rules are well known)
(Rules given by experts)
(Rules encoded using knowledge representation)
(Execution of rules by a program)
Limitation of Expert Systems
Rules may be too complex
Rules may be inexpressible
Rules may be unknown (specially for new diseases)
Here we can take help from Machine Learning, Reinforcement
Learning, Deep Learning ( No need to write rules).
Communication, perception and
Actuation
• Communication
– Traditionally AI agents (or robots) are required to
communicate with humans.
– The agent should have capability of Natural Language
Processing
– Modern NLP is data driven and ML/DL is best choice for
NLP
(Example: An agent can perform voice analysis using
Machine learning/Deep Learning)
Communication, perception and
Actuation
• Perception
– How agent perceive the environment (i.e., how agent
observe around its surrounding)
– It gives ability of vision and speech
– Modern CV (computer vision) and speech are data
driven
• Actuation
– Performing some action by an agent (i.e robot)
– Reinforcement Learning
– This is data driven now.
AI vs Data Science
Data Science AI Machine
Learning