Вы находитесь на странице: 1из 16

Introduction

Artificial Intelligence and Neural Network

Sujan Tamrakar
AI = psychology + philosophy + mathematics + linguistic
Intelligence = connects perception to action
Artificial Intelligence (AI) = tries to mimic working of human mind

• AI = seeks to understand the computations required for intelligent


behaviour & to produce computer systems that exhibit intelligence
• AI includes perception, motor control, communication using natural
language, reasoning, planning, learning & memory.
• AI is the study of how to make computers do thing at which people
are better at the moment.
• AI can be defined as a field of study to develop human level
intelligence in machine such that the machine can
• think like human, act like human, think rationally and act like rationally.
1. Acting humanly
• Alan Turing (1950) proposed a test named Turing Test to define intelligence of
a machine.
• He defined ‘intelligent behaviour’ as ability to achieve human level
performance in all cognitive tasks
• He devised a test:
Interrogate the system by human. System will pass the test if human can’t
tell whether it is human or machine he is conversing with.
To pass this test, computer would need following capabilities:
• NLP (Natural Language Processing): to interact with interrogator in natural language
(teletype)
• KR (Knowledge Representation): to store information provided before & during
interrogation
• Automated Reasoning: to use stored information to answer questions & to draw
new conclusion
• Machine Learning: to adapt to new circumstances & to detect an extrapolate
patterns.
Turing Test
2. Thinking humanly
• A.k.a. cognitive modelling approach
• Bringing together computer models from AI & experimental techniques from
psychology to try to construct precise & testable theories of workings of
human mind
• Based on experimental investigation of actual human or animals

3. Thinking rationally
• A.k.a. Laws of thought approach
• Rational thinking is right thinking
• When correct premises are given, it will produce correct conclusions
Socrates is a man.
All men are mortal.
Socrates is mortal.
4. Acting rationally
• A.k.a. Rational agent approach [agent = something that acts]
• Rational agent is one that acts so as to achieve the best outcome or when
there are uncertainty, the best expected outcome.
• Emphasis is mainly on correct inferences since machine itself knows nothing.

Applications of AI:
• Autonomous Planning & Scheduling – NASA rover exploration
• Game playing – IBM’s deep blue program for Chess, Google’s DeepMind
project for Go game
• Autonomous Control – Google car, Tesla
• Diagnosis – Medical imaging diagnosis
• Robotics – Microsurgery, military, deep-ocean exploration
• Language Understanding & Problem solving – Cross-word puzzle solver
• Expert System & Theorem proving
Objectives of AI:
• Understand human cognition – how do humans solve problems?
• Cost effective automation – replace human in intelligent tasks
• Superhuman intelligence – programs to exceed human intelligence
• Coherent discourse – communicates with people using natural language
• Learning – gather own data and its procedure
• Store information – acquire, save and retrieve information itself
Intelligent Agent
• An agent is anything
that can be viewed as
perceiving its
environment through
sensors & acting upon
that environment
through actuators.
Inference
Engine
• An ideal Rational Agent should do whatever action is expected to maximize its
performance measured on basis of evidence provided by the percept sequence
and whatever built in knowledge the agent has.

• Agent = Architecture + Agent program


• Architecture = computing device in which program will run
• Program = function that implements the agent mapping from percept to actions

• PAGE description
• Percepts
• Actions
• Goals
• Environment
• Example 1:
Agent type: Auto Taxi
Percepts: camera, microphone, GPS, speedometer
Actions: accelerate, brake, steer, produce speech, display, honk horn
Goals: safe, fast, legal, comfort, maximize profit
Environment: road, pedestrians, vehicles, traffic signal, pavement

• Example 2:
Agent type: Vacuum Cleaner agent
P:
A:
G:
E:
Types of Agent based on agent program:
1. Simple Reflex agent:
• if- then rule / reactive approach
• are really not very bright
• will not accumulate experience
• will not learn from its experience
• If the percept is not in the reflex agent's database, the agent cannot react
appropriately to the situation
• Ex: simple mercury type thermostat/ heater
2. Model based Reflex agent:
• "how the world works"
• keeping track of the part of the world it can see now
• does this by keeping an internal state that depends on what it has seen before
so it holds information on the unobserved aspects of the current state.
• Ex: This time out mars Lander after picking up its first sample, it stores this in
the internal state of the world around it so when it come across the second
same sample it passes it by and saves space for other samples.
3. Goal based agent:
• Proactive approach
• has a representation of the current state of the environment and how that
environment generally works
• These agents consider different scenarios before acting on their
environments, to see which action will probably attain a goal. This
consideration of different scenarios is called search and planning. It makes
the agent proactive, not just reactive.
• Another interesting feature of the goal-based agent is that it already has some
model of how the objects in its environment usually behave, so it can perform
searches and identify plans based on this knowledge about the world.
• Ex: Vacuum world problem
4. Utility based agent:
• add one more feature to our goal-based agent, to make it even more adaptive
• Agents so far have had a single goal.
• Agents may have to juggle conflicting goals.
• Need to optimize utility over a range of goals.
• Utility: measure of goodness (a real number).
• Combine with probability of success to get expected utility.
• a utility measure is applied to the different possible actions that can be performed in
the environment. This sophisticated planner is a utility-based agent
• utility-based agent will rate each scenario to see how well it achieves certain criteria
with regard to the production of a good outcome.
• Things like the probability of success, the resources needed to execute the scenario,
the importance of the goal to be achieved, the time it will take, might all be factored
in to the utility function calculations.
• how desirable a particular state is == utility function which maps a state to a measure
of the utility of the state
• ex: Auto taxi (Google car)

Вам также может понравиться