By Kotagiri Ramamohanarao, James Bailey (auth.), Tamás (Tom) Domonkos Gedeon, Lance Chun Che Fung (eds.)
Consider the matter of a robotic (algorithm, studying mechanism) relocating alongside the true line trying to find a specific element ? . to help the me- anism, we imagine that it may converse with an atmosphere (“Oracle”) which publications it with information about the course during which it may cross. If the surroundings is deterministic the matter is the “Deterministic element - cation challenge” which has been studied relatively completely . In its pioneering model  the matter was once provided within the atmosphere that the surroundings may possibly cost the robotic a price which was once proportional to the gap it was once from the purpose searched for. The query of getting a number of speaking robots find some extent at the line has additionally been studied [1, 2]. within the stochastic model of this challenge, we ponder the situation while the training mechanism makes an attempt to find some degree in an period with stochastic (i. e. , most likely faulty) rather than deterministic responses from the surroundings. therefore whilst it may particularly be relocating to the “right” it can be prompt to maneuver to the “left” and vice versa. except the matter being of significance in its personal correct, the stoch- tic pointlocationproblemalsohas potentialapplications insolvingoptimization difficulties. Inmanyoptimizationsolutions–forexampleinimageprocessing,p- tern acceptance and neural computing [5, nine, eleven, 12, 14, sixteen, 19], the set of rules worksits wayfromits currentsolutionto the optimalsolutionbasedoninfor- tion that it currentlyhas. A crucialquestionis oneof identifying the parameter whichtheoptimizationalgorithmshoulduse.
Read Online or Download AI 2003: Advances in Artificial Intelligence: 16th Australian Conference on AI, Perth, Australia, December 3-5, 2003. Proceedings PDF
Similar nonfiction_6 books
- Durability of Spent Nucl Fuels, Facility Components in Wet Storage (IAEA TECDOC-1012)
- Initiation à la géologie et à la topographie: à l'usage des aides-géologues, techniciens de chantiers et d'exploitation minière
- Chromic Phenomena - Technological Applns of Color Chem.
- Discrete Semiconductor Circuit Examples - ITT
- Analyzing Delayed-Fission Gamma Yields [pres. slides]
Additional resources for AI 2003: Advances in Artificial Intelligence: 16th Australian Conference on AI, Perth, Australia, December 3-5, 2003. Proceedings
Let E be an environment with the probability of correct feedback and coresponding penalty probabilities respectively. If is a new environment constructed such that its probability of correct feedback then the penalty probabilities for its two actions are The above lemma follows easily by substituting in place of in Equations (2) and (3) respectively. Thus we arrive at a dual of a given environment merely by complementing its parameter Let E be the given Deceptive environment. By definition then, we have its probability of correct response We now construct a dual of this environment with a corresponding probability of correct response Then this dual environment is Informative since, Thus if the learning autmaton can somehow determine whether a given environment is Deceptive, then Lemma 4 and 5 assure us that by interchanging the actions (or equivalently the penalties and rewards), the automaton will still be able to converge to the optimal action with as high a probability as we want.
3 Related Work Oommen  proposed and analyzed an algorithm that operates by discretizing the search space while interacting with an Informative environment. This algorithm takes advantage of the limited precision available in practical implementations to restrict the probability of choosing an action to only finitely many values from the interval [0,1). Its main drawback is that the steps are always very conservative. If the step size is increased the scheme converges faster, but the accuracy is correspondingly decreased.
As a result of this sharing of ideas, the standard of the league has steadily risen, putting pressure on the leading teams to make further leaps forward. To improve the speed of the robot, the team re-examined the parameterised walk. Perhaps there were trajectories other than a rectangular path that would allow the robot to walk faster or more smoothly? Min Sub Kim and Will Uther employed an optimisation algorithm to try to discover a better walk. During the 2003 competition, a robot could be seen pacing back forth across the field, collecting data for the optimisation.