AI, an edX reading helper.
Table of Contents
Preamble
Welcome to Artificial Intelligence!
In this course, you will learn the fundamental concepts of Artificial Intelligence (AI) and apply them to the design and implementation of intelligent agents that solve real-world AI problems, including problems in search, games, machine learning, logic, and constraint satisfaction.
We will provide a broad understanding of the basic techniques for building intelligent computer systems. Topics include the history of AI, intelligent agents, state-space problem representations, uninformed and heuristic search, game playing and adversarial search, logical agents, constraint satisfaction problems, along with techniques in machine learning and other applications of AI, such as natural language processing (NLP).
Course Level
Please note this is a Master’s/graduate level course. Expect to spend at least several hours to complete the programming assignments, although the exact amount of time will depend on your background and proficiency with coding. If you are taking this course for fun, and are not working towards a passing grade for credit, you can of course watch the lectures and answer the quizzes.
Prerequisites
Students are required to have the following prerequisites:
Linear algebra (vectors, matrices, derivatives)
Calculus
Basic probability theory
Python programming The course offers an excellent opportunity for
students to dive into Python while solving AI problems and learning
its applications. Programming assignments will be in Python.
Edit No offence, I'll try Python, with, anyway, yep; but here I'd already have my programming language of preference.
Class Schedule
Week 1.
Introduction to AI, history of AI, course logistics, and roadmap
J. McCarthy, (AI) the science and engineering of making intelligent machines. Russell & Norvig by, the study and design of intelligence agents. -> 4 schools of thought:
- Thinking humanly (how to make computers have a mind or think)
- Acting humanly (Rich, Knight, 1991)
- Thinking rationally (study the mental faculties through the use of computational models)
- Acting rationally (Poole et al, 1998) ✓ ;[adopted in this course | or how to divide intelligence systems that act rationally in order to achieve their goals]
- Applications of AI: namely:
- natural language processing
- robotics
- vision
- AI Foundation & History
- The first algorithm is believed to be the Euclides algorith to calculate the greatest common divisor (GCD).
- Overview, furthering
- agents, action or series of actions => path (to goals) which comes in different costs and depths. 2 kinds of search:
- Uninformed (not domain knowledge) techniques as breadth first search BFS, DFS (deep..) UFS (uniform cost search)
- Informed, using heuristics, information or rule. Using techniques as greedy search, A*, etc. "8 queens problem" (looking for the possible configuration so that no queen attacks another one), "route finding".
- agents, action or series of actions => path (to goals) which comes in different costs and depths. 2 kinds of search:
Road map of methods?
- some supervised learning methods: classification, K-nearest neighbourgh perception, neural networks, linear regression, boosting.
- unsupervised learning methods: clustering with k, means association rules, etc.
- constrained satisfaction: we don't care about the path but the goal itself, here the problem is generally formulated using variables instead of states, example sudoku.. CSP.
- Logical agents. PL, FOL fixed syntax. Modus Ponens (which is an inference rule), horn clauses, etc
- Reinforcement learning (designing an agent that evolves in a stochastic or certain environment), agents learn from reinforcement or delayed reward.
Week 2.
Intelligent agents, uninformed search.
- PEAS, performance, environment, actuators, sensors.
- Environment types.
- Static (vs. dynamic) the environment is changing (or any kind of agent's performance score does) while deliberating
- Discrete (vs. continuous) on defining precepts and actions.
- Type of Agents.
reflex, model-based, goal-based, utility-based,
https://mediaup.uni-potsdam.de/Player/8633 http://streamup.uni-potsdam.de/flash/69272517_hd.mp4
Week 3.
Heuristic search, greedy search, A* algorithm, stochastic search
Week 4.
Adversarial search, game playing
Week 5.
Machine Learning 1: basic concepts, linear models, K nearest neighbors, overfitting
Week 6.
Machine Learning 2: perceptrons, neural networks, naive Bayes
Week 7.
Machine Learning 3: decision trees, ensemble, logistic regression, and unsupervised learning
Week 8.
Constraint satisfaction problems
Week 9.
Markov decision processes, reinforcement learning.
Week 10.
Logical agents, propositional logic and first order logic
Week 11.
AI applications to natural language processing (NLP)
Week 12.
AI applications to vision/robotics, Course Review and
Conclusion
Assignments There will be two kinds of assignments:
Quizzes (conceptual): These test your understanding of the lectures. You may be asked to reason abstractly about the nature of an algorithm, or to perform a technique by hand on an small problem. Please read the instructions carefully, note any formatting requirements, and review your answers before hitting submit. Except for the most challenging questions, you will often only have one attempt to answer a question.
Projects (programming): These offer an excellent opportunity for you to dive into Python programming and design while solving AI problems and learning its applications. You will often be presented with a general problem and asked to come up with solutions to the problem by implementing algorithms from scratch. As mentioned above, expect to spend at least several hours to complete the programming assignments.
Grading Quizzes (20%): There will be 11 quizzes worth 2% each for a total of 20%. The lowest score will be dropped. Projects (50%): There will be 5 projects in Python worth 10% each for a total of 50%. All projects count. Final Exam (30%): There will be a final exam one week after the last lecture.