Artificial intelligence Overview by Ramya Mopidevi
Download as PPTX, PDF1 like2,190 views
The document provides an extensive overview of artificial intelligence (AI), detailing its goals, types, and enabling technologies across various fields including sociology, computer science, and biology. It discusses AI applications such as machine learning, natural language processing, computer vision, and robotics, outlining their benefits, industry trends, and the impact on sectors like healthcare and finance. Additionally, it highlights the potential challenges and ethical considerations of AI advancements, including job displacement and societal implications.
Artificial intelligence Overview by Ramya Mopidevi
- 2. Artificial Intelligence 2 Artificial intelligence aims to create intelligent machines similar to human beings that perceive their environment and take actions that maximize their chance of success at some goal. Sociology Philosophy Computer Science Psychology Neuron Science Biology Mathematic s Areas Contributing to AI Key AI Enablers Parallel and cheaper computing Innovative Algorithms Real-time Data and Big Data Knowledge Planning Reasoning Problem Solving Perception Learning Motion & Manipulation AI Goals
- 3. Types of Artificial Intelligence 3 General Intelligence Superintelligence Perform any intellectual task that a human being can Scientific Creativity, General wisdom and social skills Planning Strategy for performing an action Expert Systems Solve domain specific complex problems Machine learning Continuous learning for better prediction Speech & Voice Recognition Understand spoken language to perform an action Natural Language Processing Dialogue between human and intelligent machines Vision Systems Image processing & computer vision using learning algorithms Robotics Motion and manipulation in physical world WeakAINarrowAI FullAIAI-Complete Applied AI Strong AI
- 5. Planning 5 AI that is concerned with the decision making and realization of strategies or action sequences performed by intelligent agents, autonomous robots and unmanned vehicles when trying to achieve a goal Artificial Intelligence Planning Actions on Agents to achieve goals Goals Initial States AI Planning Systems & Techniques • The Stanford Research Institute Problem Solver (STRIPS) using Planning Domain Definition Language (PDDL) • GraphPlan • Hierarchical Task Networks • Partial Order Planning (PoP) • Preference based planning Algorithms for Planning • Classical planning • Reduction to other problems • Temporal planning • Probabilistic planning Benefits of Planning Goal directed Reducing search Resolving goal conflicts Easy error recovery
- 7. Expert Systems 7 The expert systems are the applications developed to solve complex problems in a particular domain, at the level of extra-ordinary human intelligence and expertise. Benefits of Expert Systems Improved Decision Quality High Performance Availability Speed Less error rate Steady response Domain Description Financial Decision Making Insurance companies have used expert systems to assess the risk presented by the customer and to determine a price for the insurance Design and Manufacturing Assist in the design of physical devices and processes, ranging from high-level conceptual design of abstract entities all the way to factory floor configuration of manufacturing processes. Process Monitoring and Control Analyze real-time data from physical devices with the goal of noticing anomalies, predicting trends, and controlling for both optimality and failure correction Inference Engine Strategies: Forward Chaining – “What can happen next” Backward Chaining – “Why this happened” User (May not be an expert) Domain Expert Knowledge Engineer Knowledge Base Facts (IF-THEN Rule) Inference Engine Rules & Reasoning User Interface Knowledge Expert System
- 9. Machine Learning 9 Gives the computers the ability to learn without being explicitly programmed. It continuously observes a series of actions performed over a period of time and uses this knowledge to build and enhance the predictive model for better decision making Machine Learning Input Output Algorithm Build Predictive Model Training Data Learn Algorithm Perform Tasks Feedback Supervised Machine LearningMachine Learning Supervised Algorithms that can apply what has been learned in the past to new data Unsupervised Clustering of observation data to come up with unknown pattern Deep Learn through hierarchy of simple and complex concepts Reinforcement Observe environment and make adjustments if negative for better decision making Active Learn by asking questions to increase confidence Evolutionary Learn to optimize using introduced randomness Benefits Learning from high volume data Feature learning in a short span of time Parameter optimization
- 10. Machine Learning Industry Trends 10 Top 5 Machine Learning APIs Domain Description Data Security Machine learning algorithms can look for patterns in how data in the cloud is accessed, and report anomalies that could predict security breaches. Personal Security Machine learning can speed up the screening process at public places such as airports, stadium, concerts etc. significantly to ensure safer events and eliminate false alarms Financial Trading Many prestigious trading firms use proprietary machine learning systems to predict and execute trades at high speeds and high volume. Healthcare Machine learning systems are vastly used in medical industry for predicting cancer, diabetics risk factors even before they are diagnosed Marketing Personalization Using Machine learning companies can personalize which emails a customer receives, which direct mailings or coupons, which offers they see, which products show up as “recommended” and so on, all designed to lead the consumer more reliably towards a sale. Automotive Machine learning allows smart car to learn about its owner and its environment. It will adjust the internal settings — temperature, audio, seat position, etc. — automatically based on the driver, report and even fix problems itself • Facebook ML algorithm to prevent suicide • Facebook personalized news feed • Google search engine • Google Maps • Google Assistant • Waymo - Self Driving car • Cortana Intelligence suite • Kinect Gesture Recognition • MS Word Editor to flag words Real-World Machine Learning
- 11. Deep Learning 11 Deep learning uses Artificial Neural Network with many layers to learn optimal model parameters for a feature extraction. It creates an algorithm that will automatically decide which features work best to accomplish a task. Deep learning can be trained both in supervised or unsupervised manner. • Hidden nodes form multiple layers of nonlinear processing units • These units transforms low-level input data into high- level representations of data by finding patterns at each preceding level forming complex patterns for more accurate prediction Input nodes similar to neurons that receive input signal – (images of Cat/Dog/Man/Women) Output nodes are similar to neurons that sends and output signal (Result Prediction) Benefits Inputs can be texts, images, sensors’ data and even sound Pattern recognition
- 12. Deep Learning Industry Trends 12 Application Description Colorization of Images Deep learning can be used to use the objects and their context within the photograph to color the image, much like a human operator might approach the problem. Sounds To Silent Movies A deep learning model associates the video frames with a database of pre-rerecorded sounds in order to select a sound to play that best matches what is happening in the scene. Automatic Machine Translation Text translation can be performed without any preprocessing of the sequence, allowing the algorithm to learn the dependencies between words and their mapping to a new language Object Classification in Photographs Deep learning enables classification of objects within a photograph as one of a set of previously known objects Real-World Machine Learning • Google Brain Project • DeepMind (Acquired) Deep learning algorithms are applied to the other fields like • Computer vision • Automatic speech recognition • Natural language processing • Bioinformatics
- 14. Speech and Voice Recognition 14 Automatic speech and voice recognition is computer's ability to understand and translate a spoken language and into text or perform some task associated with it. Deep Learning and Big Data are key enablers for Automatic Speech Recognition. Modern End-to-End Automatic Speech Recognition Systems • Listen, Attend and Spell (LAS) • Latent Sequence Decompositions (LSD) • Watch, Listen, Attend and Spell" (WLAS) Speech Recognition Speaker independent, hence training is not involved Voice Recognition Speaker dependent, hence software needs to be trained with unique characteristics of speaker’s voice. Typically used for biometric authentication Analog to Digital Acoustic Model Language Model Speech Engine Display Feedback Acoustic Model creates statistical representation of sounds that make up each word Language Model captures the properties of language Benefits of Speech Recognition Reduce costs Increase effectiveness Emotion analysisSpeech Recognition Process
- 15. Speech Recognition Industry Trends 15 Liv.Ai’s API – Gappi chat app available in multiple Indian languages that converts speech into text. Application Description Hands-free Navigation Ask destination distance and time to reach on GPS connected digital maps Automated Identification Create a ‘voiceprint’ based on specific text such as ‘Name’ and ‘Account Number’ which is stored against the individual’s record. so when they next call, they can simply say their name and the person is put straight through to a customer service representative Removing IVR menus ‘intelligent call steering’ (ICS) does not involve any ‘button pushing’. The system simply asks the customer what they want (in their words, not yours) and then transfers them to the most suitable resource to handle their call. • Google • Microsoft • IBM • Baidu - Deep Speech 2, • Apple • Amazon • Nuance • SoundHound • IflyTek Top Speech Recognition Companies
- 17. Natural Language Processing 17 Natural language processing refers to AI method of communicating with intelligent system permitting a human-computer dialogue in a conversational, day-to-day natural language such as English. Natural Language Understanding (NLU) • Mapping the natural language input into useful representations • Analyzing different aspects of the language Natural Language Generation (NLG) • Text planning: retrieving relevant content from knowledge base • Sentence planning: Choosing required words, forming phrases and setting tone of the sentence • Text Realization: Mapping sentence plan into sentence structure NLP is difficult because of ambiguities associated with inconsistencies in human natural language Benefits Fast return on value Real-time analysis Handles any format of information parsing Translating Generating Input Phonology: Interprets sound within and across words Morphology: Breakdown words into morphemes Lexical: Assign meanings to individual words Syntactic: Check if sentence is grammatically correct Semantic: Performs disambiguation of words Discourse: Makes connection between sentences Pragmatic: Look contextual and situational meanings 1 2 3 4 5 6 7
- 18. Natural Language Processing Industry Trends 18 • Expect Labs • SwiftKey • NetBase • FiscalNote • Kelvu Top Speech Recognition Companies Application Description NLP in Text Prediction Text prediction technology designed to significantly boost the accuracy, fluency and speed of text entry on mobile and computing devices by learning from users’ writing style and predict favorite words, phrases and emojis NLP in Social Media Analysis Software to the data from the social web to apply social media sentiment analysis using NLP technologies NLP in Predicting Government Legislation Product for analyzing political, legal, and regulatory information using NLP and machine learning NLP in eCommerce Analysis System to help merchants improve the shopping experience on their site, and increase conversions and revenue
- 20. Computer Vision 20 Technology through which a computer can see by extracting, analyzing, and comprehending useful information from a single image or an array of images through algorithms Optical Character Recognition convert image into editable text Estimating Position – position of tumor in the body Object Recognition – Parking number plate recognition. Facial recognition expression based customer sentiment Devices Required: • Camera • Processer • Software • Display for monitoring Benefits No limitation like as human perception Easy to work with the devices (mount, remove, replace and upgrade) The Goal of CV is to emulate the striking perceptual capability of human eyes and brains or even to surpass and assist humans The other vision systems are: • Image Processing • Machine Vision
- 21. Computer Vision Industry Trends 21 FB: New computer vision algorithms can “read” images and videos to the blind and display over 2 billion translated stories every day • Cognex • Datalogic • IVISYS • Microscan • National Instruments • Optotune • ProPhotnix • Sensory • USS Vision • ViDi Systems Application Description Object Tracking in video Use the color of an object to track its trajectory as it moves in the video. Plant classification Color histograms and machine learning to classify the plants Image search Feature-based learning and recognition algorithm to re-rank the outputs from a traditional keyword-based image search engine Sharpening, blur, and noise removal Image processing for the enhancement of images through the use of sharpening and noise removal operations Facial animation A parameterized 3D model of shape and appearance (surface texture) can be used directly to track a person’s facial motions and to animate a different character with these same motions and expressions Top Computer Vision Companies
- 23. Robotics 23 Robots are aimed at manipulating the objects by perceiving, picking, moving, modifying the physical properties of object, destroying it, or to have an effect thereby freeing manpower from doing repetitive functions without getting bored, distracted, or exhausted. Sensors (vision and tactile sensors) Actuators Electric, piezo and ultrasonic motors Pneumatic air muscles Muscle wires Mechanical construction Electrical Components Computer Program Form, or shape designed to accomplish a particular task Power and control the machinery Determine what, when and how a robot does something Legged Wheeled Legged & Wheeled Track Skip/Slid SwimmingDronesStationary
- 24. Robotics Industry Trends 24 • Alphabet, Inc. (Google) • Amazon • iRobot • Lockheed Martin • Samsung • Toyota • Foxxconn Technology Group • ABB Robotics • EPSON Robotics • FANUC Robotics • KUKA Robotics • Rethink Robotics • Yamaha Robotic • Yaskawa Robotics, Top Robotic Companies by Robotics Business Review Industry Application Industrial Robots are used for handling material, cutting, welding, color coating, drilling, polishing, etc. Military Autonomous robots can reach inaccessible and hazardous zones during war. Healthcare Robots are capable of carrying out hundreds of clinical tests simultaneously, rehabilitating permanently disabled people, and performing complex surgeries such as brain tumors. Exploration the robot rock climbers used for space exploration, underwater drones used for ocean exploration are to name a few. Entertainment Disney’s engineers have created hundreds of robots for movie making Automotive Autonomous or self driving car are now becoming a reality
- 26. Industries that are most impacted due to AI • Dynamic & data intensive markets • Financial adviser dependent • Customer service support • Security/privacy/fraud sensitive • Highly regulated • Innovators from lab to real-world • Venturing into diverse industries • Nurture innovation and skills • Investing in startups & incubators • Personalize user experience • Highly dependent on underwriters & agents • Clients preference for self-service • Data driven claims & settlements • Increase in personal & commercial IoT devices • Self-driving cars • Shift from ‘owned’ to ‘rent a service’ • Usage based Auto-insurance • Personalized driver assist features • Increase in IoT & Telematics • Preventive & early diagnosis • Personal care for old & disabled • Shortage of nurses & surgeons • Large clinical data repository • Too many medical publications • Health & fitness tracker data • Simple, routine & heavy lifting work to reduce production time • Work in danger zones • Supply chain & inventory mgmt.. • Equipment & fleet monitoring • On-demand delivery • Online reviews influence on shoppers • Personalized shopping experience • Cross product recommendations • Marketing campaigns, offers & discounts Disruptive business are shifting their focus from Information Technology to AI-Powered Technology. IT/ITES companies that already have huge tech talent needs to focus on early adoption of AI Technology by: Launching interactive AI courses to train their domain SMEs Nurturing in-house innovation though hackathons Investing in right AI tech start-ups Collaborating with Universities for R&D in AI & emerging technologies Technology Giants Insurance AutomotiveBFSi Healthcare Retail & e-commerce IT/ITESManufacturing & Logistics Intelligent search engine; Personalized recommendations & newsfeeds; voice powered assistant & chatbots; AI platforms & products Automated financial advisors & brokers; Chatbots; AI user authentication; Smart wallets; Credit scoring models; Intuitive intelligence for fraud prevention Robo-underwriters & Robo-agents; Automated underwriting & pricing models; Sophisticated NLP & vision systems algorithms for settlements Robo-taxi; Connected cars; Voice powered assistant; AI vision systems & NLP models for accident prevention & remote monitoring Robot assisted surgeries & patient care & monitoring; Designing treatment plans; Online consultation; Prediction models for drug creation Robots for assembling & packaging; drones for equipment monitoring & ML for predictive maintenance; Collaborative warehouse robots Intelligent recommendations engine; Chatbots, Voice & AR/VR based shopping experience; customer sentiment; AI models for store operations; Drone warehouse InfluencersDisruptionsInfluencersDisruptions 26 US Federal Report says Artificial Intelligence could automate 47% of jobs
- 27. A few Disruptive AI Innovations 27 Amazon - “People who bought this also bought…”, “recommended for you” on the other sites Facebook – News feed, “Add a Friend” Google - Google Translate, Search Engine Gaming - The world’s best Checkers, Chess, Scrabble, Backgammon, and Othello players Amazon, Google, Facebook, IBM, and Microsoft have established a non-profit partnership to formulate best practices on artificial intelligence technologies, advance the public's understanding, and to serve as a platform about artificial intelligence.
- 29. Which Artificial Intelligence is our Future? 29 Artificial Narrow Intelligence Some real-world issues Unemployment due to skill replacement through automation 2010 Flash Crash – AI failure let to stock market brief plummet, taking $1 trillion of market value To avoid legal battle Amazon Echo hands over data to police in a murder case. Apple & FBI battle over unlocking terrorist’s FBI ANI AGI Artificial General Intelligence Human Level Machine Intelligence – a machine that can perform any intellectual task that a human being can Ex: A machine that can create another machine, build sky scrappers etc. Threat to Privacy, Safety, Human Dignity and potential devaluation to humanity by artificial moral agents ASI Artificial Super Intelligence The Law of Accelerating Returns Technological Singularity - Superintelligence will abruptly trigger runaway technological growth, resulting in unfathomable changes to human civilization to extent that human existence is under question. We are Here! Self-improvement of Algorithms, content & Hardware Recursive self- improvement causing Intelligence explosion
- 30. 30 Thank You