Illinois Genetic Algorithms Laboratory Department of General Engineering

Illinois Genetic Algorithms Laboratory Department of General Engineering University of Illinois at Urbana-Champaign Urbana, IL 61801. DISCUS: Moving from Decision Support Systems to Innovation Support Systems David E. Goldberg, Michael Welge, & Xavier Llora NCSA/ALG + Illi. GAL University of Illinois at Urbana-Champaign deg@uiuc. edu

Innovation This & Innovation That l l The business world is abuzz with “innovation. ” Popular books tell companies how to get it. But little scientific understanding of what it is. UIUC research changing that.

Decision Support & Knowledge Management versus Innovation Support l l l Decision support systems help evaluate enumerated alternatives. Knowledge management helps manage that which is known. Is it possible to create new class of innovation support system to systematically permit organizations to use IT to support pervasive and persistent innovation?

Collaboration + Key Ideas = Opportunity l Previous collaboration of ALG + Illi. GAL – – – l Confluence of key ideas – – – l Applications-ready GA theory MOGAs for D 2 K & the real world Interactive genetic algorithms Interactive GAs Human-based GA (Kosorukoff & Goldberg, 2002) Chance discovery & data-text mining DISCUS: Distributed Innovation and Scalable Collaboration in Uncertain Settings

Overview l l l l 3 elements research from Illi. GAL 4 trips to the South Farms 2 trips to Japan The innovation connection The key problem: interactive superficiality Key. Graphs as aid to reflection Key elements of DISCUS

Three Elements from TB l Illi. GAL has studied principled – – – l l l Genetic algorithm design theory Genetic algorithm competence Genetic algorithm efficiency Design theory permits analysis w/o tears. Competence = solve hard problems, quickly, reliably, and accurately O(l 2). Efficiency takes tractable subquadratic solutions to practicality.

Four Trips to the South Farms l Collaboration had blossomed with ALG & Prof. Minsker on – – l Carrying principled design theory to practice Multiobjective selection to D 2 K & practice GBML and HBGAs to D 2 K Interactive GAs Keys for the current project: – – HBGAs Interactive GAs

Interactive & Human-Based Genetic Algorithms l l Interactive GAs replace machine eval with human eval Human-based GAs replace ops & eval with human: www. 3 form. com Figure : Actual photo of simulated criminal (above). Evolved image from witness using Faceprints (below).

Two Trips to Japan l l Visited Tsukuba University, Graduate School of Systems Management, December 2001 – January 2002. Met Key. Graph Inventor & Chance Discovery Proponent, Yukio Osawa. Did Tutorial with Dr. Osawa August 2002. Finally understood importance of topic & relation to GAs.

Modes of Innovation l GA as model of innovation – – l Chance discovery – l Kaizen = selection + mutation Discontinuous change = selection + crossover Low probability events linked to matters of importance Keygraphs as one computational embodiment of chance discovery. http: //www-doi. ge. uiuc. edu

Innovation This & Innovation That l l l The business world is abuzz with “innovation. ” Popular books tell companies how to get it. But little scientific understanding of what it is.

Selection+Recombination = Innovation l Combine notions to form ideas. l “It takes two to invent anything. The one makes up combinations; the other chooses, recognizes what he wishes and what is important to him in the mass of the things which the former has imparted to him. ” P. Valéry

Scalable Solutions on Hard Problems l l l Time to first global averaged over five runs. Subquadratic versus quintic. Compares favorably to hillclimbing, too (Muhlenbein, 1992).

Key. Graph Example: Japanese Breakfast Figure: Key. Graph (Ohsawa, 2002) shows two clusters of food preferences for Japanese breakfast eaters. The chance discovery of rare use of vitamins was viewed as a marketing opportunity by food companies.

Key Problem & Notion l l Human-based GAs interesting, but suffer from interactive superficiality. Key. Graphs have been used to gain insight into text data, but usually batch mode of processing. l l Combine interactivity of HBGAs and insight promotion of Key. Graphs. Boost everything with competent efficient GAs and IEC at population outskirts.

Summary l l l l 3 imports from Illi. GAL 4 trips to South Farms 2 trips to Japan The innovation connection Key problem: interactive superficiality Possible solution: interactive collaboration with reflection boosted by Key. Graphs Larger framework with competent & interactive GEC.

More Information l Visit Illi. GAL web site. l http: //www-illigal. ge. uiuc. edu/ l l Recent book: Goldberg, D. E. (2002). The Design of Innovation. Boston, MA: Kluwer Academic. Attend GECCO-2003, Chicago, July 12 -16, 2003, www. isgec. org/GECCO-2003/.

People Involved With DISCUS l llinois Genetic Algorithms Lab - Department of General Engineering - University of Illinois at Urbana Champaign (Illi. GAL-UIUC) – l Automated Learning Group - National Center for Supercomputing Applications (ALG-NCSA) – l Hideyuki Takagi Graduate School of Systems Management - University of Tsukuba – l Barbara Minsker, Abhishek Singh, Meghna Babbar Kyushu Institute of Design - Department of Art and Information Design – l Ali Yassine, Miao Zhuang Minsker Research Group - Department of Civil and Environmental Engineering - University of Illinois at Urbana-Champaign – l Alan Craig Deparment of General Engineering - National Center for Supercomputing Applications – l Tim Wentling, Andrew Wadsworth, Luigi Marini, Raj Barnerjee Data Mining and Visualization Division - National Center for Supercomputing Applications – l Michael Welge, Loretta Auvil, Duane Searsmith, Bei Yu Knowledge and Learning System Group - National Center for Supercomputing Applications – l David E. Goldberg, Xavier Llorà, Kei Ohnishi, Tian Li Yu, Martin Butz, Antonio Gonzales Yukio Ohsawa VIAS (the Visualization Information Archival/Retrieval Service)