ADVANCED ARTIFICIAL INTELLIGENCE Syllabus Topic 1 - Modern AI Topic 2 - Machine Learning Topic 3 - Deep Learning Topic 4 - Big Data Topic 5 - Classification Topic 6 - Regression Topic 7 - Clustering Topic 8 - Convolutional Neural Network Lecturer Assoc. Prof Dr Azlan bin Mohd Zain UTM Big Data Centre Universiti Teknologi Malaysia Skudai Johor. [emailprotected] my
What is intelligence? • • • Many definitions of Intelligence ! The ability to learn The ability to recognize problems The abilty to saolve problems The ability to learn from experiences The ability to adapt to the surrounding environment
What is Artificial Intelligence? • AI is the simulation of human intelligence processes by machines, especially computer systems. • The tasks that constitute AI include: ü ü Problem Solving Knowledge Representation Decision Making Actuation and Perception
Classical AI • Machines will be capable of doing any work human can do.
Classical AI • The goal of classical AI was to explicitly represent human knowledge using facts and rules. • That is you program a machine with some rules and when you input a query, it gives the answer based on the rules. • Taking the example of the calculator, it is programmed with the rules of mathematical operations and when you input 2+3, it returns 5.
Classical AI • Classical AI worked very well in certain situation, but it still suffered from a major problem, that of rules. • What if there are lots of rules or if the rules are not well defined. • It becomes difficult to codify or specify the rules explicitly. • Example, if you want a machine to identify a cat, what rules would you specify to be able to identify it?
Modern AI (2008+) • Modern AI is capable to handle many of the weaknesses of classical AI. • Unlike classical AI, modern AI doesn’t need people to explicitly specify rules for it. • It is capable of learning rules on it own. • It can infer and extract the rules and patterns required to get from the input to the output.
Modern AI • Modern AI is in short a powerful method of pattern recognition and in many ways mimics how people learn. • Example, if you want a machine to be able to identify cats, you give it tons of cat pictures and have it understand what are the common features of cats. • Then whenever it sees another cat it is able to match its features and identify it as one.
Modern AI • The rise in modern AI can be attributed to three factors: ü Data Volume ü Statistical models ü Computing power
Modern AI • Data, and not just any volume of data but huge volume of data. • With a small volume of data, modern AI becomes very narrow. • You might have lots of data, and is extremely important for modern AI.
Modern AI • The rise in computing power due to better CPUs and GPUs means that the huge data volume aren’t simply stored in databases but can now be used. • This allows for applying new statistical models on the huge data volume using better computing resources to extract valuable information and patterns. • We can in short call this as Data Science.
Data Science • The term “Data science” was introduced in early 2008 by two data team leads from Linkedin & Facebook. (DJ Patel & Jeff Hammerbacher). • A look at the evolution of the data landscape, AI becomes the tool that helps data science get results and the solutions for specific problems.
How Data Science intersects the AI world? 14
Data Science vs. the world of AI • Some people say Data Science and Artificial Intelligence are synonyms. • Some say that one is a subset of the another and some say that these terms are completely unrelated.
AI - Data Science related terminology • Data Science and Artificial Intelligence typically have different objectives. • AI and Data Science overlaps at two cores of sub-knowledge, Data and Machine Learning. 16
What is Data Science? • Data Science is the science of collecting, storing, processing, describing and modelling data. 17
1. Collecting data • Collecting data depends on the question which is raised and the business environment. 18
Collecting data - Example 1 • Assume the data scientist is working for an e -commerce company. • A question the data scientist may be interested in, is Which items do customers buy? • Here the data already exists within the organisation and the data scientist must have knowledge of accessing the database using SQL etc. 19
Collecting data - Example 2 • Assume the Data scientist is working for a political party. A question the data scientist may be interested in, is What is people’s opinion about the new agenda of the party? • Here the data exists (in from of tweets, Facebook posts…) but not within the organisation. • So the data scientist must have web crawling (scraping) and other skills apart from programming skills. 20
Collecting data - Example 3 • Assume the Data scientist is working with farmers. • A question the data scientist may be interested in, is Effect of type of seed, fertiliser, irrigation on yield? • Here the data is not available and the data scientist needs to design experiments to collect data using Statistics etc. 21
Collecting data - Needed skills • Knowledge of Databases • Intermediate Programming • Knowledge of Statistics 22
2. Storing data • Data Warehouse ü used for analytical and reporting purposes. ü comes under OLAP (Online Analytical Processing). ü capable of both relational and multidimensional data. • Database ü used for transactional purposes. ü comes under OLTP (Online Transaction Processing). ü capable of storing relational data. 23
Storing data 24
Unstructured data • Unstructured data includes text, data, video and speech etc. • This type of data has 5 V’s characterize Big Data stored in Data Lakes. ü ü ü High volume High variety High velocity High value High veracity 25
5 V’s Big Data 26
Storing data - Needed skills • • • Programming Knowledge of relational databases Knowledge of NOSQL databases Knowledge of Data Warehouses Knowledge of Data Lakes(Hadoop) 27
3. Processing data • Data wrangling ü Integrating data from external resources which are in different format into database • Data cleansing ü Filling missing values, correct spelling errors, identify and remove outliers and standardize keyword tags 28
Processing data • Data scaling • • ü Scaling included converting data in kilometers to miles or rupees to dollars etc. Data normalizing ü Normalizing includes ensuring data has zero mean, unit variance. Data stardardising ü Standardizing data includes having values between zero and one. 29
Processing data - Needed skills • • Programming Map Reduce(Hadoop) SQL and NOSQL databases Basic Statistics 30
4. Describing data • Visualising data ü Understanding how the data looks for effective understanding is Data Visualisation. ü Examples include grouped bar charts and scatter plots. • Summarising Data ü Summarising data helps to answer questions related to it. Inferential Statistics and Descriptive Statistics help in Describing data in a better way. 31
Describing data - Needed skills • • • Statistics Excel Python R Tableau 32
5. Modeling Data • The stage where most people consider interesting. As many people call it “where the magic happens”. • Two types of modeling, statistical modeling and algorithmic modeling. 33
Statistical Modeling • In simple terms, statistical modeling is a simplified, mathematicallyformalized way to approximate reality. • Make predictions from the approximation. • The statistical model is the mathematical equation that is used. 34
Algorithmic Modeling • Focuses on building functions that deals with high dimensional data. • The goals include: ü Estimating the function using data and optimisation techniques. ü Given a new input, Predict the output. 35
Modeling data - Needed skills • • • Probability theory Inferential statistics Calculus Optimization algorithms Machine learning & deep learning Python packages & frameworks (Numpy, Scikit-learn, TF, Py. Torch, Keras etc. ) 36
Summary • The rise in Modern AI can be attributed to three factors, they are data volume, statistical models and computing power • Data Science is a comprehensive process that involves collecting, storing, processing, describing and modelling data. • On the other hand, AI is the implementation of a predictive model to forecast future events. • Data Science comprises of various statistical techniques whereas AI makes use of computer algorithms. 37
References • https: //dimensionless. in/understanding-differentcomponents-roles-in-data-science/ • https: //medium. com/ai-in-plain-english/data-science-vsartificial-intelligence-vs-machine-learning-vs-deep-learning 50 d 3718 d 51 e 5 • https: //medium. com/datadriveninvestor/data-sciencedemystified-what-is-data-science-ef 031 f 6 d 0 b 6 d • https: //www. educba. com/data-warehouse-vs-database/ • https: //towardsdatascience. com/ai-the-past-the-present-and -the-future-f 00 f 399 b 2 c 51
The main research topics in AI include: problem solving, reasoning, planning, natural language understanding, computer vision, automatic programming, machine learning, and so on. Of course, these topics are closely related with each other.
ARTIFICIAL INTELLIGENCE IN MODERN SOCIETY. 4. Artificial intelligence is fundamentally intelligence behavior displayed by machines instead of humans. People who have studied computer science are needed in creating AI because they should observe how the machine reacts to certain variables.
According to the current system of classification, there are four primary AI types: reactive, limited memory, theory of mind, and self-aware.
- Reactive Machines.
- Limited Memory.
- Theory of Mind.
- Artificial Narrow Intelligence (ANI)
- Artificial General Intelligence.
- Artificial Super Intelligence (ASI)
Tech professionals specializing in AI and ML is super high, and hence, there is a crunch of employees in the market. Salary for the CSE employee starts from INR 3 lakh per annum, while that of a professional with an AI and ML background gets a job with INR 6 lakh per annum.
Extremely promising. In fact, there are more artificial intelligence jobs than skilled professionals to fill them, and the AI world has shown no signs of slowing down, so the demand is very high.
If John McCarthy, the father of AI, were to coin a new phrase for "artificial intelligence" today, he would probably use "computational intelligence." McCarthy is not just the father of AI, he is also the inventor of the Lisp (list processing) language.
Prominent examples of AI software used in everyday life include voice assistants, image recognition for face unlock in mobile phones, and ML-based financial fraud detection.
Artificial intelligence is generally divided into two types – narrow (or weak) AI and general AI, also known as AGI or strong AI.
Artificial intelligence is widely used to provide personalised recommendations to people, based for example on their previous searches and purchases or other online behaviour. AI is hugely important in commerce: optimising products, planning inventory, logistics etc.
In AI research, math is essential. It's necessary to dissect models, invent new algorithms and write papers.
Learning AI is not an easy task, especially if you're not a programmer, but it's imperative to learn at least some AI. It can be done by all. Courses range from basic understanding to full-blown master's degrees in it.
Linear Algebra is the primary mathematical computation tool in Artificial Intelligence and in many other areas of Science and Engineering. With this field, you need to understand 4 primary mathematical objects and their properties: Scalars — a single number (can be real or natural).
Yes, if you're looking to pursue a career in artificial intelligence and machine learning, a little coding is necessary.