BMEE407L - Artificial Intelligence Module 1 PDF

BMEE407L - Artificial Intelligence Dr. R. Mohan, Ph.D., Professor SMEC, VIT Chennai  BMEE407L Artificial Intelligence Pre-requisite: BMAT202L & BMAT202P Probability and Statistics Pre-requisite: BMAT101L & BMAT101P Calculus  BMEE407L - Artific...

BMEE407L - Artificial Intelligence Dr. R. Mohan, Ph.D., Professor SMEC, VIT Chennai  BMEE407L Artificial Intelligence Pre-requisite: BMAT202L & BMAT202P Probability and Statistics Pre-requisite: BMAT101L & BMAT101P Calculus  BMEE407L - Artificial Intelligence Course Objectives 1.To provide basic understanding on Artificial Intelligence with its sub-sets. 2.To impart knowledge of search algorithm, logics, reasoning and uncertainty. 3.To introduce the basic concepts of machine learning and its application in mechanical engineering.  BMEE407L - Artificial Intelligence Course Outcome At the end of the course, the student will be able to 1.Translate the characteristics of artificial intelligence and its sub-sets. 2.Implement appropriate algorithm for problem solving by searching. 3.Construct the logical agents and familiar in the application of fuzzy in AI. 4.Design the decision making algorithm with the reasoning of uncertainties. 5.Develop machine learning programs based on supervised, unsupervised and reinforcement learning. 6.Experiment the benefit of neural network in deep learning. BMEE407L - Artificial Intelligence  BMEE407L - Artificial Intelligence Text Books 1. Russell S, Norvig P, Artificial Intelligence - A Modern Approach, 2021, 4th edition, Prentice Hall. Agenda Item 65/46 - Annexure - 42 2. Ivan Vasilev, Advanced Deep Learning with Python: Design and implement advanced next-generation AI solutions using TensorFlow and PyTorch, 2019, 1st edition, Packt Publishing Ltd. Reference Books 1. Bishop C. M, Pattern Recognition and Machine Learning, 2011, 2nd edition, Springer. 2. Nilsson N.J, Artificial Intelligence: A New Synthesis, 1998, 1st edition, Morgan Kaufmann.  Course Plan Module 1 & 2 CAT-1 (as per schedule) Module 5 & 6 CAT-2 (As per schedule) Module 3, 4, 7 & Contemporary Discussion  FAT Exam (As per schedule) Digital Assignment1: Before CAT1 2 IBM certificates or equivalent (AWS / Python courses and so on). Based on the platform on which the mini project (Assignment3) to be carried out. Digital Assignment2: Before CAT2 2 IBM certificates or equivalent (AWS / Python courses and so on). Based on the platform on which the mini project (Assignment3) to be carried out. Digital Assignment3: Before FAT Mini AI Project based on Assignment1 & Assignemnt2 learning as well as class room learning and self learning  Question Paper Pattern CAT-I & CAT-II (50 Marks) Part A (5 x 10 = 50 Marks) FAT Exam (100 Marks) Part A (7 x 10 = 70 Marks) Part B (2 x 15 = 30 Marks) Note: Any change in question pattern will be communicated in advance  Faculty Details Dr. R. Mohan, Ph.D., Professor SMEC VIT Chennai Cabin Location: AB1 – 406A, 3D Printing Lab Contact#: 9884216335 Email ID: [email protected]  Module:1 Foundation of AI (4 Hours) Introduction – Foundations of AI – Evolution of AI – Intelligent Agents: Agents and environments, Concept of rationality, structure of agents – Structure of Knowledge based system - Risks and Benefits of AI.  Introduction Artificial intelligence is a field of science concerned with building computers and machines that can reason, learn, and act in such a way that would normally require human intelligence or that involves data whose scale exceeds what humans can analyze. With the help of AI, machines are programmed to understand, think, learn, behave like humans and mimic human behavior. AI allows computers to process large amounts of data and identify patterns in them, gradually learn from their own experience, adapt to the set parameters, and perform human-like tasks, such as  Four Approaches in Defining AI Systems that act like humans Systems that think like humans Systems that think rationally Systems that act rationally  1. Acting humanly: The Turing test approach The art of creating machines that perform functions that require intelligence when performed by people i.e., making computers to act like humans. Example : Turing Test Provides satisfactory operational definition for intelligence. Provided by conducting Turing Test The Turing Test is a method of inquiry in artificial intelligence (AI) for determining whether or not a computer is capable of thinking like a human being. The test is named after Alan Turing, the founder of the Turing Test  1. Acting humanly: The Turing test approach The system requires these abilities to pass the test Natural language processing for communication with human Woman, Machine & Judge. Knowledge representation Which one’s the computer? Pass the test ? to store information effectively if the interrogator cannot & efficiently A tell Automated reasoning if there is a computer or to retrieve & answer questions using the stored information B a human at the other end. Machine learning to adapt to new circumstances  1. Acting humanly: The Turing test approach Turing viewed the physical simulation of a person as unnecessary to demonstrate intelligence. However, other researchers have proposed a total Turing test, which requires interaction with objects and people in the real world. To pass the total Turing test, a robot will need computer vision and speech recognition to perceive the world 2. Thinking humanly: The cognitive modeling approach Making computers to think like humans Goal is to build systems that function internally in some way similar to human mind Also called cognitive modeling approach We can learn about human thought in three ways Example GPS (General introspection—trying to catch our own thoughts Problem Solver) as they go by; comparing the reasoning psychological experiments—observing a person in steps of the program and human solving the same action; problems. brain imaging—observing the brain in action Precise theory of the mind =>becomes possible to express the theory as a computer program. If the program’s input–output behavior matches corresponding human behavior, that is evidence that some of the program’s mechanisms could also be operating in humans  3. Thinking rationally: The “laws of thought” approach Thinking “Right Thing” : Making computers to think the “Right Thing” Relies on logic (to make inferences) rather than human to measure correctness. Logic: provide a precise notation for statements about all kinds of things in the world and the relations between them. Syllogism: Provide patterns for argument structures Always give correct conclusion given correct premises. For example, Premise: John is a human and all humans are mortal Conclusion: John is mortal Can be done using logics. Example Propositional and predicate logic.  3. Thinking rationally: The “laws of thought” approach Two obstacles: Laws of Thought of Approach It’s not easy to take informal knowledge and state it in the formal terms required by logical notation, particularly when the knowledge is less than 100% certain. Being able to solve a problem “in principle” and doing so “in practice” are very different. i.e., 1. Informal Knowledge is not precise. 2. Difficult to model uncertainty 3. Theory and practice cannot be put together.  4. Acting rationally: The rational agent approach Doing “Right Thing” A rational agent is one that acts so as to achieve the best outcome or, when there is uncertainty, the best expected outcome Design of rational agent Advantages More general than laws of thought approach Concentrates on scientific development Limited Rationality acting appropriately when there is not enough time to do all computation AI has focused on the study and construction of agents that do the right thing. What counts as the right thing is defined by the objective that we provide to the agent.  AI Careers & Examples of artificial intelligence include: Career Description Job Outlook Smart assistants like Siri and Alexa Path Pandora and Netflix, which provide Big Data Find meaningful patterns in data by looking at the The need for skilled AI personalized song and entertainment Analyst past to help make predictions about the future. professionals spans nearly recommendations User every industry, including: Chatbots Experience Work with products to help customers understand their function and can use them easily. Robotic vacuum cleaners (UX) Understand how people use equipment and how Financial services Designer/Deve computer scientists can apply that understanding Self-driving vehicles to produce more advanced software. Healthcare loper Facial recognition software Technology Natural Language Explore the connection between human language Media and computational systems; this includes working Processing on projects like chatbots and virtual assistants. Marketing Engineer Government and military Researcher Work with computer science and AI research Discover ways to advance AI technology National security Expert in applied math, machine learning, deep Research learning, and computational stats. Expected to have an advanced degree in computer science or IoT-enabled systems Scientist an advanced degree in a related field supported by experience. Agriculture Software Develop programs in which AI tools function. The Gaming Engineer role may also be referred to as a Programmer or Artificial Intelligence Developer. Retail AI Engineer Build AI models from scratch and help product managers and stakeholders understand results. programming Top 5 skills required for AI additional skills for AI Data Mining Finding anomalies, patterns, etc. within large data languages jobs: professionals: and Analysis sets to predict outcomes. Python Communication skills Solid knowledge of applied Machine C/C++ Knowledge and experience mathematics and algorithms Learning Using data to design, build and manage ML software applications. MATLAB with Python specifically (in Problem-solving skills Engineer general, proficiency in Industry knowledge Data Scientist Collect, analyze and interpret data sets. programming language) Management and leadership skills Business Digital marketing goals and Machine learning Intelligence Analyze complex data sets to identify business and market trends  AI Careers & Job Outlook Companies Currently Hiring AI Positions In general, tech companies (both software A recent search for “artificial intelligence” job and hardware) dominate the list of openings on LinkedIn revealed thousands and companies that are hiring AI professionals. thousands of results at a wide variety of But a quick search on any reputable job companies. Here is a sample of some of the listing site will give you a list of positions positions we found. that span a variety of industries. Here is a Wells Fargo — Sr. Conversational AI Content sample of some of the top companies that Strategists are hiring for these types of AI roles: Nike — Data Scientist, Experience Research & Analytics Deloitte Amazon Web Services — Machine Learning Amazon Engineer Accenture Apple — AI/ML Software Engineer H&R Block Spotify — Research Scientist – Language IBM Technologies PwC Microsoft — Senior Researcher Fidelity Investments As you can see from the list above, there are PayPal many different types of positions within artificial Major League Baseball intelligence. Some of the most common AI-related Harvard Business School job titles, courtesy of Glassdoor, include: IKEA History of AI Introduction History of the various AI areas. Introduction  The Foundations of Artificial Intelligence The foundations of Artificial Intelligence (AI) encompass various interdisciplinary principles and methodologies that enable the creation and functioning of intelligent systems. Ethics & Philosophy: Ethical Considerations, Philosophical Questions Mathematics: Linear Algebra, Probability and Statistics, Calculus, Discrete Mathematics Cognitive Science: Psychology, Neuroscience Computer engineering: Algorithms and Data Structures, Complexity Theory, Programming Languages Control theory and cybernetics: Behavioural Control, Optimization and Adaptation, System Stability, Controller Design, Self-Regulating Systems, Complex Systems Modelling Computer Vision: Image Processing, Object Recognition, Segmentation Robotics: Perception, Planning, Control Linguistics: syntax, semantics, phonetics, phonology, morphology, pragmatics, and discourse analysis Machine Learning: Supervised Learning, Unsupervised Learning, Reinforcement Learning, Deep Learning Natural Language Processing (NLP): Syntax and Semantics, Language Models,  The Foundations of Artificial Intelligence Key Concepts and Techniques Knowledge Representation: Methods for encoding information about the world in a form that AI systems can utilize. Search Algorithms: Techniques for navigating through problem spaces to find solutions (e.g., A*, Dijkstra's algorithm). Logic and Reasoning: Using formal logic to derive conclusions from known facts (e.g., propositional logic, first-order logic). Bayesian Networks: Probabilistic graphical models for representing and reasoning about uncertainty. Markov Decision Processes (MDPs): Models for decision- making in situations where outcomes are partly random and partly under the control of a decision maker. Applications of Artificial Intelligence  Intelligent Agent An agent is anything that can be viewed as perceiving its environment through sensors and acting upon that  The Structure of Agents The job of AI is to design an agent program that implements the agent function - the mapping from percepts to actions We assume this program will run on some sort of computing device with physical sensors and actuator - we call this the agent architecture gent = architecture + program The Structure of Agents The Structure of Agents  Intelligent Agent Simple reflex agent Model-based reflex agents Goal-based agents Utility-based agents Learning agents Simple Reflex Agent Simple Reflex Agent Simple Reflex Agent Simple Reflex Agent Model Based Reflex Agent Model Based Reflex Agent  Model Based Reflex Agent smart home heating, ventilation, and air conditioning (HVAC) system. Environmental Model: Includes the layout of the house, thermal properties of rooms, and typical occupancy patterns. Weather Model: Considers current and forecasted weather conditions. User Preferences: Includes preferred temperature and https://www.fibaro.com/en/smart-home-in-use/ humidity settings.  Model Based Reflex Agent mart Irrigation in Agriculture Environmental Model: Includes data on soil types, crop water requirements, and weather patterns. State Model: Maintains current soil moisture levels, weather conditions, and the status of irrigation https://intellias.com/smart-irrigation-in- agriculture/ equipment. https://www.youtube.com/watch?v=WWINyVj2MIo Goal-based agents  Goal-based agents utonomous delivery drone Environmental Model: Includes maps of the delivery area, no-fly zones, and known obstacles. State Model: Maintains the drone's current position, battery level, and package status. Goal: Deliver the package to the specified address efficiently and safely. https://www.youtube.com/watch?v=Ih3oN7J-v5w https://www.youtube.com/watch? v=qs2VeyzRMzQ Utility-based agents  Utility-based agents smart home heating, ventilation, and air conditioning (HVAC) system. Utility Function: Comfort Level: Weighted utility for maintaining a comfortable temperature and humidity. Energy Efficiency: Weighted utility for minimizing energy usage. Cost: Weighted utility for minimizing the cost based on real-time electricity An autonomous Utility rates. Function: drone is tasked with delivering packages to various Deliverylocations. Time: Ensuring the package reaches its destination as quickly as possible. Safety: Avoiding obstacles and adverse weather conditions to minimize the risk of crashes. Battery Life: Managing its flight path to conserve battery and ensure it has enough power to return to the base if needed. Customer Preferences: Delivering the package within a specific time window if Learning agents  Learning agents ersonalized Recommendation System Netflix's AI recommendation engine analyzes massive amounts of data, including viewing habits, ratings, searches, and time spent on the platform, to curate personalized content recommendations for each viewer. https://research.netflix.com/research-area/ machine-learning  Concept of rationality and Performance Concept of rationality A rational agent is an agent whose acts try to maximize some performance measure. An agent should strive to "do the right thing", based on what it can perceive and the actions it can perform. The right action is the one that will cause the agent to be most successful Performance measures An objective criterion for success of an agent's behavior E.g., performance measure of a vacuum-cleaner agent could be amount of dirt cleaned up, amount of time taken, amount of electricity consumed, amount of noise generated, etc. The rationality of an agent depends on four things the performance measure defining the agent's degree of success the percept sequence, the sequence of all the things perceived by the agent the agent's prior knowledge of the environment the actions that the agent can perform  Definition of Rational agents Rational Agent: For each possible percept sequence, a rational agent should select an action that is expected to maximize its performance measure, given the evidence provided by the percept sequence and whatever built-in knowledge the agent has. A rational agent should be autonomous  Task Environment To design a rational agent, we need to specify a task environment a problem specification for which the agent is a solution PEAS: to specify a task environment Performance measure Environment Actuators Sensors  Task Environment PEAS: Specifying an automated taxi driver Performance measure: ? Environment: ? Actuators: ? Sensors: ?  Task Environment PEAS: Specifying an automated taxi driver Performance measure: safe, fast, legal, comfortable, maximize profits Environment: roads, other traffic, pedestrians, customers Actuators: steering, accelerator, brake, signal, horn Sensors: cameras, sonar, speedometer, GPS  Task Environment Performance Agent Type Environment Actuators Sensors Measure Automated Safe, fast, legal, Roads, Steering Cameras, sonar, Taxi Driver comfortable trip, traffic, accelerator, speedometer, maximize profits pedestrian brake, GPS, odometer, customers signal, accelerometer horn, display engine sensors, keyboard Medical Healthy patient, Patient, Screen Keyboard (entry of diagnosis minimize costs, hospital, staff display symptoms, system lawsuits (question findings, patient's tests, answers) diagnoses treatment referrals) Part-Picking Percentage of Conveyor belt Jointed arm Camera, joint Robot parts in correct with parts, bins and hand angle sensors bin Interactive Maximize Set of students Screen Keyboard  Task Environment Performance Agent Type Measure Environment Actuators Sensors robot amount of goals soccer match cameras, sonar or soccer legs scored field infrared player Satellite Display Correct Image Downlink from Image categorizati Color pixel arrays Categorization satellite Analysis on of scene Valves, Temperature, Refinery Maximum purity, Refinery pumps, pressure, chemical controller safety operators heaters, sensors displays minimize energy Vacuum consumption, Left, Right, Sensors to identify two squares Agent maximize dirt Suck, NoOp the dirt pick up  Properties of task environments 1.Fully observable vs. Partially observable Fully observable – If an agent’s sensors give it access to the complete state of the environment at each point in time then the environment is effectively and fully observable. There will be no portion of the environment that is hidden for the agent. Real-life Example 1: While running a car on the road ( Environment ), The driver ( Agent ) is able to see road conditions, signboard and pedestrians on the road at a given time and drive accordingly. So Road is a fully observable environment for a driver while driving the car. Example 2 : A chess playing system is an example of a system that  Properties of task environments Partially observable The agent is not familiar with the complete environment at a given time. Real-life Example: Playing card games is a perfect example of a partially-observable environment where a player is not aware of the card in the opponent’s hand  Properties of task environments 2.Deterministic vs. stochastic Deterministic next state of the environment is completely determined by the current state and the actions executed by the agent, then the environment is deterministic, otherwise, it is Stochastic. – Real-life Example: The traffic signal is a deterministic environment where the next signal is known for a pedestrian (Agent)  Properties of task environments stochastic The Stochastic environment is the opposite of a deterministic environment. The next state is totally unpredictable for the agent. Real-life Example 1: The radio station is a stochastic environment where the listener is not aware about the next song. Example 2:Taxi driving is clearly stochastic in this sense, because one can never predict the behavior of traffic exactly;  Properties of task environments 3.Episodic vs. sequential Episodic An episodic environment means that subsequent episodes do not depend on what actions occurred in previous episodes. Real-life Example 1: A support bot (agent) answer to a question and then answer to another question and so on. So each question-answer is a single episode. Example 2: Pick and Place robot, which is used to detect defective parts from the conveyor belts.  Properties of task environments Sequential In a sequential environment, the agent engages in a series of connected episodes. In sequential environments, on the other hand, the current decision could affect all future decisions. Real-life Example: Checkers- Where the previous move can affect all the following moves. Medical Diagnosis –Diagnosis diseases is wrong, giving medicine, taking all test is going to be wrong.  Properties of task environments 4.Static vs. dynamic Dynamics A dynamic environment is always changing over time Example: the number of people in the street Static Environment is not changed over time Example: Crossword Puzzle  Properties of task environments 5.Discrete vs. continuous Discrete It consists of a finite number of states and agents have a finite number of actions. Example: A chess game comes under discrete environment as there is a finite number of moves that can be performed.  Properties of task environments Continuous It consists of a infinite number of states and agents have a infinite number of actions. Example: Taxi driving is a continuous state and continuous-time problem.  Properties of task environments 6.Single agent VS. multiagent Single agent: An environment is explored by a single agent. All actions are performed by a single agent in the environment. Real-life Example: Brushing a teeth Multiagent: If two or more agents are taking actions in the environment, it is known as a multi-agent environment. Real-life Example :Playing Cards  Properties of task environments 7.Known vs. unknown The agent’s knows about the complete environment and the outcomes for all actions. example: solitaire card games - If the environment is unknown, the agent will have to learn how it works in order to make good decisions.( example: new video game). Task Environment Observable Determini Episodic Static Discrete Agent stic Crossword puzzle Fully Determinist Sequenti Static Discrete Single ic al Chess with a Fully Stochastic Sequenti Semi Discrete Multi clock al Poker Partially Stochastic Sequenti Static Discrete Multi al Backgammon Fully Stochastic Sequenti Static Discrete Multi al Taxi driving Partially Stochastic Sequenti Dynami Continuou Multi al c s Medical diagnosis Partially Stochastic Sequenti Dynami Continuou Single al c s Image-analysis Fully Determinist Episodic Semi Continuou Single ic s Part-picking Partially Stochastic Episodic Dynami Continuou Single robot c s Refinery Partially Stochastic Sequenti Dynami Continuou Single controller al c s  Structure of Knowledge Based System One of the most famous KBSs was MYCIN, a program for medical diagnosis. The real-world facts were represented as a simple assertion  Structure of Knowledge Based System Acquisition and maintenance: There is no need for a programmer to maintain the facts. The domain experts can define and maintain the rules themselves. Function: Facilitates the process of gathering knowledge from experts, documents, databases, or other sources. Tools and Methods: May include interviews, questionnaires, data mining techniques, or machine learning algorithms to capture and update knowledge. Explaining: Existing facts can be used to infer a new conclusion, and results can be explained for usage purposes. A simple example can be to follow a series of inferences towards a diagnosis, and then use these facts to clarify the diagnosis. Importance: Provides users with insights into the reasoning process of  Structure of Knowledge Based System Reasoning: Decoupling knowledge from the processing of that knowledge enabled general-purpose inference engines to be developed. That system can bring new solutions never seen before by developers, which follow from a data set stored in the knowledgebase. Deductive Reasoning: Derives conclusions from general rules. Inductive Reasoning: Infers general rules from specific cases. Abductive Reasoning: Infers the best explanation for a given set of Knowledgebase observations. A KBS uses its knowledgebase to store all the knowledge data. The gathered knowledge from experts, either human or artificial, is first symbolized and then stored Content: Contains domain-specific facts, rules, and relationships. Representation: Knowledge can be represented using various methods such as: Rules: If-then statements. Frames: Data structures for dividing knowledge into substructures by representing stereotyped situations.  Structure of Knowledge Based System Inference Engine Role: The brain of the KBS, It applies logical rules to the facts stored in the knowledgebase to deduce new knowledge. Techniques: Common methods include: Forward Chaining: Starts with known facts and applies inference rules to extract more data until a goal is reached. Backward Chaining: Starts with goals and works backward to determine what facts must be asserted to achieve those goals. An IE operates in three states: i.e. matching rules, selection rules and execution rules. The IE filters all the content of the data that are satisfied by the rules. The content items are considered execution candidates. In the selection rules, some selection strategies are applied to further filter the rules for execution. In the broader AI context, the term heuristics is used for these selection strategies. Finally, the data are executed in the third state of the IE, where the right-hand sides of the stored rules are changed, or further processed, by any input from outside the IE, either from  Risks and Benefits of AI As AI plays an increasingly important role in the economic, social, scientific, medical, financial, and military spheres. The risks and benefits—that it can bring The potential for AI and robotics to free humanity from menial repetitive work and to dramatically increase the production of goods and services could presage an era of peace and plenty. The capacity to accelerate scientific research could result in cures for disease and solutions for climate change and resource shortages. Customized Experiences: AI algorithms can provide personalized recommendations in areas like e-commerce, streaming services, and social media. Healthcare: AI can tailor medical treatments to individual patients based on their genetic information and health data.  Risks and Benefits of AI As AI plays an increasingly important role in the economic, social, scientific, medical, financial, and military spheres. The risks and benefits—that it can bring Lethal autonomous weapons: Surveillance and persuasion Biased decision making: Impact on employment: Safety-critical applications: Cyber security: