CAP 5636 - Advanced Artificial Intelligence

• understand the search and decision making techniques used in modern artificial intelligence
• apply artificial intelligence techniques in their own code
• understand the societal and ethical implications of artificial intelligence

• Stuart Russel and Peter Norvig, Artificial Intelligence - A Modern Approach, 4rd edition

• Only full grades will be used based on the points obtained. A for 90 and above, B for 80-89, C for 70-79, F for lower than 70.
• Points awarded: Midterm 1: 20 points, Midterm 2: 20 points, Homeworks/Projects: 30 points total, Final exam 30 points.
• Some midterms, exams and homeworks will have bonus points, but no curve will be applied.
• The exams will be administered through ProctorHub, and are open book, open notes.
• Make up exams will be given only in justified cases.

• In case of an emergency, dial 911 for assistance.
• Every UCF classroom contains an emergency procedure guide posted on a wall near the door. Please make a note of the guide's physical location and consider reviewing the online version.
• If there is a medical emergency during class, we may need to access a first aid kit or AED (Automated External Defibrillator). To learn where those items are located in this building, see the link (click on link from menu on left).
• To stay informed about emergency situations, sign up to receive UCF text alerts by going to my.ucf.edu and logging in. Click on "Student Self Service" located on the left side of the screen in the tool bar, scroll down to the blue "Personal Information" heading on your Student Center screen, click on "UCF Alert," fill out the information, including your e-mail address, cell phone number, and cell phone provider, click "Apply" to save the changes, and then click "OK."

Please take the time to familiarize yourself with Zoom by visiting the UCF Zoom Guides at . You may choose to use Zoom on your mobile device (phone or tablet).

Things to Know About Zoom:

• You must sign in to my Zoom session using your UCF NID and password.
• The Zoom sessions are recorded.
• Improper classroom behavior is not tolerated within Zoom sessions and may result in a referral to the Office of Student Conduct.
• You can contact Webcourses@UCF Support at if you have any technical issues accessing Zoom.

Please see the Zoom housekeeping slide developed by UCF instructional designer, Trudy Trail-Constant.

To protect members of our community, everyone is required to wear a facial covering inside all common spaces including classrooms (https://policies.ucf.edu/documents/PolicyEmergencyCOVIDReturnPolicy.pdf. Students who choose not to wear facial coverings will be asked to leave the classroom by the instructor. If they refuse to leave the classroom or put on a facial covering, they may be considered disruptive (please see the Golden Rule for student behavior expectations). Faculty have the right to cancel class if the safety and well-being of class members are in jeopardy. Students will be responsible for the material that would have been covered in class as provided by the instructor.

Notifications in Case of Changes to Course Modality

Depending on the course of the pandemic during the semester, the university may make changes to the way classes are offered. If that happens, please look for announcements or messages in Webcourses@UCF or Knights email about changes specific to this course.

COVID-19 and Illness Notification

Students who believe they may have a COVID-19 diagnosis should contact UCF Student Health Services (407-823-2509) so proper contact tracing procedures can take place.

Students should not come to campus if they are ill, are experiencing any symptoms of COVID-19, have tested positive for COVID, or if anyone living in their residence has tested positive or is sick with COVID-19 symptoms. CDC guidance for COVID-19 symptoms is located here: (https://www.cdc.gov/coronavirus/2019-ncov/symptoms-testing/symptoms.html)

Students should contact their instructor(s) as soon as possible if they miss class for any illness reason to discuss reasonable adjustments that might need to be made. When possible, students should contact their instructor(s) before missing class.

In Case of Faculty Illness

Depending on the course of the pandemic during the semester, the university may make changes to the way classes are offered. If that happens, please look for announcements or messages in Webcourses@UCF or Knights email about changes specific to this course.

Course Accessibility and Disability COVID-19 Supplemental Statement

Accommodations may need to be added or adjusted should this course shift from an on-campus to a remote format. Students with disabilities should speak with their instructor and should contact sas@ucf.edu to discuss specific accommodations for this or other courses.

• Motivating AI. Dangers of AI and AGI.
• Early history
• Expert systems

• Neural networks
• The two intellectual traditions of logic vs neural networks
• A melting pot of other ideas
• The agent view of AI

• Reflex agents
• Search problems
• Depth first and breadth first search
• Uniform cost search

• Informed search methods
• Heuristics
• Greedy search
• A* search
• Graph search

• Types of games
• Adversarial search, minimax
• The problem of depth
• Evaluation functions
• Alpha Beta pruning

• Expectimax search
• Refresher about probabilities
• Utilities and rationality

• Defining MDPs: policies and utilities
• Optimal policy, value of state, value of Q-state

• Policy iteration

• Reinforcement learning as a twist on MDPs

• Model-based and model-free learning
• Temporal difference learning

• Exploration vs. exploitation, regret
• Generalization across states
• Policy search

• Random variables
• Joint and marginal distributions, conditional distribution

• Product rule, chain rule, Bayes' rule
• Inference
• Independence

• Markov chains
• Conditional independence
• Stationary distributions

• Hidden Markov models
• Example: robot localization

• Most likely explanation
• Speech recognition

• Particle filters
• Robot localization with particle filters
• Dynamic Bayes nets

• Classification
• Model-based classification
• Naive Bayes
• Spam filter example
• Generalization and overfitting
• Parameter estimation

• Classification and machine learning cont'd

• History and impact

• History and impact
• Machine learning background
• Loss functions: squared, cross-entropy, softmax
• Optimization, stochastic gradient descent
• Backpropagation

• Feedforward networks
• Stochastic gradient descent

• Convolutions
• Convolutional filters in neural networks
• Pooling layers

• General definitions of intelligence
• Cognitive architectures
• Superintelligence and singularity