Foundation of Al and Ml (4351601) PDF

Foundation of AI and ML (4351601) Fundamental of - H. P. Jagad Lecturer(IT) Sir BPTI Bhavnagar http://hpjagad.blogspot.com 1 ...

Foundation of AI and ML (4351601) Fundamental of - H. P. Jagad Lecturer(IT) Sir BPTI Bhavnagar http://hpjagad.blogspot.com 1   Artificial = Man-made  Intelligence = Thinking power Man-made thinking power  Intelligence is the ability of a system to calculate, reason, Introduction learn from experience, store and retrieve information from To memory, Solve problems, use natural language fluently, classify and adapt new situations. Artificial Data Information Knowledge Intelligence Intelligence  It is a branch of computer science by which we can create intelligent machines which can ✓behave like a human, ✓think like humans, ✓able to make decisions. Unit-1  Why AI? 2  Intelligence composed of..  Problem Solving PLRPL  Learning − gaining knowledge or skill by studying, practicing or experiencing.  Reasoning- provide basis for judgement & making decisions  Perception − It is the process of acquiring, interpreting, selecting sensory information.  Linguistic Intelligence − Ability to use, speak, and write the verbal and written language. Unit-1 3 Unit-1 4  11. Online Ads-Network 1. Google Maps 12. Banking and Finance 2. Face Detection and recognition 13. Smart Home devices 3. Text Editors and Autocorrect 14. Security and 4. Chatbots Surveillance 5. E-Payments 15. Smart Keyboard App Example of AI 6. Search and Recommendation 16. Smart Speaker algorithms 17. E-Commerce 7. Digital Assistant 18. Smart Email Apps 8. Social media 19. Music and Media 9. Healthcare Streaming Service 10.Gaming 20. Space Exploration Unit-1 5 Unit-1 6  Based on Based on Capability Functionality Types of AI Unit-1 7   As a human being we are following some moral principles which are related to good manners and behaviors known as ethics.  Certain ethics related to AI are called AI ethics.  Ethics means what is right, Unethical means what is wrong.  While developing car’s algorithm, developer goes through dilemmas.  Morality will be transferred from developer to machine.  Assume that car has hit the boy, who is responsible? AI Ethics Unit-1 8  1. Data Privacy  Transparent System: - While collecting data the purpose and the detailed guide about data should be known to the users.  Freedom of leaving system & Data Deletion: - After using such system if user want to leave the system, the freedom should be given to the users. When the user leaves the system, his data should be deleted.  Right of data collection: - Without collecting data, AI system can not take right decisions for the user. 2. Social Impact AI Ethics  Increasing Inequalities: - Only upper-class people who can afford AI-enabled devices have the opportunity to access it and people below the poverty line don’t have access to it.  Unemployment  Negative Adoptions: The negative minded community can Unit-1 misuse this technology. 9  3. AI Bias & Fairness 4. Transparency :- AI models arrive at conclusions/decisions  It means favoring someone without providing any  Mostly all virtual assistant explanations as to how they AI Ethics has female voice. were reached. That is called  When software used to Black Box problem. So, predict future criminals transparency is required. showed bias against black 5. Security people. 6. Responsibility: - It should be  Mitigate bias & ensure clearly mentioned that who is fairness. responsible for decisions. 7. Human Control Bias Unit-1 10   High Cost - H/W & S/W are very costly.  Computing Time  Increase dependency on machines  Limited understanding of context  Limited creativity Limitations  Lack of Emotion of AI  Lack of common sense  Data Privacy  Bias & Fairness  Ethical Dilemmas Unit-1 11  AI Areas Unit-1 12  1. Expert Systems  It Is computer program that is designed to solve complex problems and to provide decision-making ability like a human expert.  User Interface: With it, system interacts with the user, take query & pass it.  Inference Engine: With the help of it, the system extracts the knowledge from the knowledge base using the reasoning and inference rules according to the user queries. Core component Application:  Knowledge Base: It is a type of storage that stores knowledge acquired from the  Medical Diagnosis: MYCIN (Diagnose blood different experts of the particular domain. clotting diseases), PXDES (Determine Type Factual Knowledge: Based on facts & rules of lung cancer), CaDeT (Detect cancer). Heuristic Knowledge: Based on practice,  Financial Analysis, Technical Support, ability to guess, evaluation & experiences. Quality Control Unit-1 13  Characteristics of Expert System  Rule base System: - Collection of “if-then” rules  Explanation Mechanism  Learning Capability  Diagnostic & Troubleshooting 1.  Decision Support: - System can assist humans. Not used to replace human expert. Expert  Domain Specific: - Every system are designed for specific Systems domain. Limitations  Need an extensive knowledge base  Difficulties to handle incomplete information  Performance is based on expert’s knowledge stored in knowledge base. Unit-1 14   It is subfield of AI used to enable computer to understand, interpret & generate human language in a valuable way.  Processing of Natural Language is required 2. ✓when you want an intelligent system like robot to perform as per your instructions, Natural ✓when you want to hear decision from a dialogue based Language expert system. Processing  The input and output of an NLP system can be: Speech and Written Text  Example- Google Assistant, Amazon Alexa, Samsung Bixby, Microsoft Cortana, Apple Siri Unit-1 15  Applications of NLP  Text understanding  Language Modelling: - Probabilities of word sequences 2.  Sentiment Analysis: - Determine sentiments or emotions Natural  Machine Translation: - Google translate Language  Speech Recognition: - Virtual assistant Processing  Question Answering  Language Generation: - ChatGPT  Conversational AI: - Chatbot IRCTC-Disha  Text Classification Unit-1 16  3. Neural Networks  Neuron or nerve cell is fundamental unit of brain that send message to all over body.  Neurons are interconnected in a layered structure that resembles human brain called Neural Network.  The human brain is composed of 86 billion neurons. They are connected to other thousand cells by Axons.  Inputs from sensory organs are accepted by dendrites. These inputs create electric impulses, which quickly travel through the neural network.  A neuron can then send the message to other neuron to handle the issue or does not send it forward.  Artificial Neural Networks are composed of multiple artificial neurons called nodes or units. Nodes imitate biological neurons of human brain.  Neurons are connected by linksUnit-1 & interact with each other. 17   Organized into layers: Input Layer, Hidden Layer, Output Layer  Nodes can take input data and perform operations on the data. The result of these operations is passed to other neurons.  Output at each node is called its activation or node value. Each link is associated with weight. 3.  Ex- Pattern recognition systems such as face recognition, character recognition, handwriting recognition. Neural Networks Unit-1 18   Robots are the artificial agents acting in real world.  Robotics is a branch of AI, which is composed of Electrical, Mechanical and Computer Science for designing, 4. Robotics construction, and application of robots.  Have mechanical construction, form, or shape designed to perform a particular task.  Have electrical components which power and control the machinery.  contains computer program that determines what, when and how a robot does something.  Ex- Agriculture, Autonomous vehicles, Forensics, security, and surveillance, Industrial quality inspection, medical imagery Unit-1 19   'Fuzzy’ means not clear.  Sometimes, we cannot decide in real life that the 5. given problem or statement is either true or false. Fuzzy Logic  At that time, this concept provides many values Systems between the true and false and gives the flexibility to find the best solution to that problem.  In the Boolean system, only two possibilities - 0 & 1, ✓1 - absolute truth value ✓0 - absolute false value.  But in the fuzzy system, there are multiple possibilities present between the 0 & 1, which are partially false and partially true.  Fuzzy Logic Systems (FLS): - Branch of AI & Mathematics that produce acceptable definite output for incomplete, ambiguous or inaccurate input. Unit-1 20   Fuzzy Sets: Set of varying degrees between 0 and 1.  Membership functions: - Assign degree of membership between 0 & 1. Quantify linguistic term. They characterize fuzziness, whether the elements in fuzzy sets are discrete or 5. continuous. Fuzzy Logic  Linguistic variable: - Describe imprecise or qualitative information like, “Cold”, “Warm” or “Hot”. Systems  Fuzzy logic Operators: - AND, OR and NOT  Fuzzy Rules: - Express relationship between i/p & o/p variables Key using linguistic terms & fuzzy logic operators. Components  Fuzzy Inference: - Process of applying fuzzy rules to i/p data to determine appropriate o/p values. Example  Photocopiers, Television, Microwave Ovens, Refrigerators, Toasters, Vacuum Cleaners, Washing Machines, AC, Heaters. Unit-1 21  1. Rule Base: - Stores set of rules and If- 5. Fuzzy Logic Systems* Then conditions given by the experts which are used for controlling the decision-making systems. 2. Fuzzification: - Transforms the inputs [crisp number-inputs which are measured by the sensors] into fuzzy steps. It divides the input signals into following five states: Large Positive (LP), Medium Positive (MP), Small (S), Medium Negative (MN), Large negative (LN). 3. Inference Engine: - Allows users to find the matching degree between the current fuzzy input and the rules. After the matching degree, this system determines which rule is to be added 4. Defuzzification: - It takes the fuzzy set according to the given input field. inputs generated by the Inference Engine, When all rules are fired, then they are and then transforms them into a crisp value. combined for developing the control The crisp value is a type of value which is actions. acceptable by the user. Unit-1 22   Astronomy: - Complex universe problems  Gaming: - Chess  Data Security: - AEG bot, AI2 Platform  Social Media: - Social Media sites such as Facebook, Twitter, and Snapchat  Travel & Transport: - AI-powered chatbots Applications  Automotive Industry: TeslaBot of AI  Robotics: - Erica and Sophia  Entertainment: - Netflix or Amazon  Agriculture: - agriculture robotics, solid and crop monitoring, predictive analysis  E-commerce: - Discover associated products  Education: -AI chatbot communicate with students as a teaching assistant Unit-1 23   Medical Image Analysis: - ✓Diagnostic Imaging, ✓Pathology, ✓Dermatology  Disease Diagnosis and Prediction  Drug Discovery & Development AI in  Telemedicine & Remote Monitoring Healthcare  Healthcare Management  Genomic & DNA Sequencing  Radiation Therapy Planning  Drug delivery & Medication Management  Mental Health Support Unit-1 24   Algorithmic Trading  Risk assessment & Fraud Detection  Customer Service  Personalized Financial Advice  Credit Scoring  Investment Portfolio Management AI in  Anti Money Laundering (AML) & Know Your Finance Customer(KYC)  Trading & Investment Strategy  Voice & Speech Recognition  Alternative Data Analysis  Chatbot for Financial Planning Unit-1 25   Predictive maintenance – Predict when m/c is likely to fail.  Quality Control  Process Optimization  Supply Chain Management  Robotics & Automation  Customization & Personalization AI in  Inventory Management Manufacturing  Production planning & Scheduling  Fault detection & correction  Employee Safety  Quality forecasting  Material tracing  Cost optimization Unit-1 26