Simulation Default Policy Backpropagation G Agent State (s) Reward (r) Environment (e.g. Maze) Action () 01 0 2 13 14 1 15
0 2 01 02 13 14 15 0 P 01 02 13 14 15 0 Fitness value
Reward Convolution Action Pooling Chapter 3: Playing Games Win Experience Player Non-Player Motivation Games as AI testbeds, AI that challenges players, Simulation-based testing
Motivation Playing roles that humans would not (want to) play, Game balancing Examples Examples Board Game AI (TD-Gammon, Chinook, Deep Blue, AlphaGo, Libratus), Jeopardy! (Watson), StarCraft Rubber banding Motivation Simulation-based testing, Game demonstrations Motivation Believable and human-like agents
Examples Examples Game Turing Tests (2kBot Prize/Mario),Persona Modelling AI that: acts as an adversary, provides assistance, is emotively expressive, tells a story, Imperfect Information Perfect Information Observability Pac-Man Atari 2600
Checkers Chess Go Ms Pac-Man Ludo Monopoly Backgammon Time Granularity Super Mario Bros Halo StarCraft Battleship Scrabble Poker Real-Time Turn-Based
Perform Square Step Experience Agnostic Experience Driven Player (Experience) Super Mario Bros (Pedersen et al., 2010) Sonancia (Lopes et al., 2015) SpeedTree (IDV, 2002) StarCraft Maps (Togelius et al., 2013) Autonomous
Sentient Sketchbook (Liapis et al., 2013) Garden of Eden Creation Kit (Bethesda, 2009) Ropossum (Shaker et al., 2013) Tanagra (Smith et al., 2010) Mixed-Initiative Designer (Initiative) Chapter 5: Modeling Players Input Computational (Player) Model Output
Gameplay Model-Based [Top-Down] Objective (Psychology, Cognitive Science, Game Studies, ) Free Response vs. Forced Response Context Player Profile First Person vs. Third Person Discrete vs. Continuous Numerical (Interval)
Regression Time-Discrete vs. Time-Continuous Nominal (Classes) Classification Pre vs. During vs. Post Ordinal (Ranks) Preference Learning No Output Unsupervised Learning (Clustering, Frequent Pattern Mining) Model Free [Bottom-Up]
Portion Safe Hit Points Exit Safe Exit Exit Safe Chapter 6: Game AI Panorama DO (Process) WHAT (Context) FOR WHO (End User) GAME AI AREA
Designer Model Content Player AI Researcher Generate Behavior Producer / Publisher Generate Content (Assisted) Model Players (Behavior, Experience) Generate Content (Autonomously) Play Games
(Win [NPC], Experience [NPC]) Play Games (Win [PC], Experience [PC]) Model Players (Behavior) Player Interaction Model Players Experience Behavior Game Generate Content Autonomously Content
NPCs Play Games [as NPC] Win Experience Play Games (as PC or NPC) For the (Game) To Win Experience Model Players Experience Behavior Generate Content Assisted Autonomously
Play Games (as PC or NPC) For the (Game) To Win Experience Model Players Experience Behavior Generate Content Assisted Autonomously Play Games (as PC or NPC) For the (Game) To Win Experience Model Players Experience
Behavior Generate Content Assisted Autonomously Play Games (as PC or NPC) For the (Game) To Win Experience Model Players Experience Behavior Generate Content Assisted Autonomously
Play Games (as PC or NPC) For the (Game) To Win Experience Model Players Experience Behavior Generate Content Assisted Autonomously
auf Basis von SAP Netweaver04 Torsten Niemietz, CIO Nordzucker AG, Braunschweig Stand 06| 2005 Agenda Agenda Nordzucker in Zahlen Konzern Kennzahlen zur IT SAP TV Das Nordzucker Portal-Projekt TIP (Transaktions- und Informations-Portal) Überblick - Auswahlprozess, Budget, Hardware Szenario III: Content-Management...
Underfill theory: vessels in the renal circulation constricted due to dilation of blood vessels in splanchnic circulation (supplies intestines), mediated by factors released by liver disease.. Decrease in "effective" volume of blood sensed by the juxtaglomerular apparatus > renin secretion...
Marking - Literacy KS1 Fantastic phrases/sentences highlighted green Errors (grammar/poor vocabulary choices) highlighted in yellow _____ = incorrect letter formation (correct formation modelled at bottom of piece of work and practice letters provided) Double tick = especially good use of...
Note that End-of-Year Reporting from ESA is not final, and final homes treated reported in the Annual Report may be slightly higher. "ESA Eligible" is the total number of SoCalGas customers that are income-eligible to participate in ESA in 2017...
Julia Zachary (Wesleyan University) Ian Steer (NASA/IPAC Extragalactic Database) Facilitators: Brandon Lawton (STScI) & James Manning (Universe of Learning) Science Briefing. January 6, 2017. Cosmic Double Whammy: Black-Hole Blast Followed by Galaxy Cluster Collision.
Impersonal "Se" Ever seen a sign that says, Se habla español. What does that mean? It can be translated, "We speak Spanish," "Spanish is spoken," "One speaks Spanish," etc. The truth is that there's not a literal translation for it....
Introduction to the concept of Evaluation. BASIC DIFFERENCES . BETWEEN MONITORING AND EVALUATION. Monitoring. Are project activities being implemented according to plan? Who is benefitting from the project? Are the expected outcomes/output being developed? Is the amount of benefits being...
e.g., read the chapter on chemical equations and construct a useable theory of them; use them in the next chapter on acids and bases. Every acid has a conjugate base, formed by removing a proton from the acid.
Ready to download the document? Go ahead and hit continue!