Скачать презентацию Optimizing Online Auction Bidding Strategies with Genetic Programming Скачать презентацию Optimizing Online Auction Bidding Strategies with Genetic Programming

5ac2f9edd34b208c7b964957b0de514e.ppt

  • Количество слайдов: 14

Optimizing Online Auction Bidding Strategies with Genetic Programming Ekaterina “Kate” Smorodkina Optimizing Online Auction Bidding Strategies with Genetic Programming Ekaterina “Kate” Smorodkina

Why Optimize Bidding Strategies? § § § Popularity of online auctions Limited resources (i. Why Optimize Bidding Strategies? § § § Popularity of online auctions Limited resources (i. e. $$$) Bidding on multiple items increases the complexity of the decision making process Increasing number of buyers and auction listings Difficulty in predicting the behavior of other buyers

Overview § Research questions and problem § § § definition Online auction overview and Overview § Research questions and problem § § § definition Online auction overview and auction simulation Strategy representation Fitness evaluation Evolving agents Experiments Future work

Research Questions § § Is it possible to come up with one all -purpose Research Questions § § Is it possible to come up with one all -purpose bidding strategy for various online auction scenarios? How successful is genetic programming in evolving bidding strategies for online auctions?

Online Auctions Overview § § § Limited time Starting bid Email notification Identical items Online Auctions Overview § § § Limited time Starting bid Email notification Identical items for sale Unrestricted bidding

Online Auction Simulation § Item listings are randomly created § § § with a Online Auction Simulation § Item listings are randomly created § § § with a starting bid and time limit Agents are created with random lists of items to buy Concurrent bidding. Retail price on each item is known Agents know if they no longer hold the highest bid on an item Agents are not allowed to go over their account balance

Strategy Representation § Expression trees § Binary operators: +, -, /, *, %, max, Strategy Representation § Expression trees § Binary operators: +, -, /, *, %, max, min § The size of the trees is controlled by two parameters: the branching limit and the depth limit § Large trees take longer time to compute a bid § Input parameters to the expression tree

Input parameters to the expression § Account balance § Retail price § Current bid Input parameters to the expression § Account balance § Retail price § Current bid § Number of items on the list § Number of items missing § Sum of the retail prices on the missing items § The highest bid among all instances of the item § The lowest bid among all instances of the item

Agent Fitness Evaluation Maximize the number of items obtained. § Maximize discount. § N Agent Fitness Evaluation Maximize the number of items obtained. § Maximize discount. § N – number of items obtained M – number of items on the list R. P – retail price H. B – highest bid §

Evolution Cycle Modified Initialize Bid Compete Evaluate Select Bid Reproduce Evolution Cycle Modified Initialize Bid Compete Evaluate Select Bid Reproduce

Evolving Agents § Proportional Selection § Recombination § Subtree crossover § Mutation § Competition Evolving Agents § Proportional Selection § Recombination § Subtree crossover § Mutation § Competition § Elitist strategy § Termination § Fitness convergence § Mutation rate adjustment

Experiments to Perform § Change the environment after each auction round § Number and Experiments to Perform § Change the environment after each auction round § Number and characteristics of items in the auction § Agents’ lists and their initial account balance § Fitness standard as a way to measure the success of the experiment

Results § None yet Results § None yet

Future Work § Finish this project § Expand the types of operators in the Future Work § Finish this project § Expand the types of operators in the expression trees § Expand input parameters to the expression trees § Create seller agents