• View
• Edit
• History
• Print

We gradually go down from the easiest to the hardest, expanding the amount of space to explore and thus the resource to involve.

This API is the core yet might require interactions (especially during early phase through the diagnosis questions) thus the need for a business side to it, the equivalent of an "human mediated API". (at StMaur SNCF office station on a yellow note 20/04/2010)

Dynamic associative table

cf ~utopiah/desktop\ before\ cloudification/hohop.txt (01/04/2010)

Statistics of assigned algorithms to problems

History of what has been tried before to refine and/or re-order the associative table

Possibility space to explore

1. from small
   1. existing and easy to automate

• solved problems with proved optimality
• solved problems considered optimal
• solved problems through heuristics but considered non-optimal
• ... and everything in between
• GA, GP, GE
• ES
  • CMA-ES
  • NES
  • xNES
• supervised GA
• hybrid GA
• unsupervised GA
• total randomness

1. to large
   1. difficult to define and presumably in a very large amount of possibilities

Relying not solely on algorithms (AImatrix and Services) but also datatsets (DATAmatrix and Services) and more generally integrated in order to provide simulations (innovation chain).

CEC 2005 IEEE Congress on Evolutionary Computation, Proceedings (see COmparing Continuous Optimisers (COCO) platform for systematic and sound comparisons of real-parameter global optimizers for updates)

• BLX-GL50 Garcia-Martinez and Lozano (Hybrid Real-Coded...)
• BLX-MA Molina et al. (Adaptive Local Search...)
• CoEVO Posik (Real-Parameter Optimization...)
• DE Ronkkonen et al. (Real-Parameter Optimization...)
• DMS-L-PSO Liang and Suganthan (Dynamic Multi-Swarm...)
• EDA Yuan and Gallagher (Experimental Results...)
• G-CMA-ES Auger and Hansen (A Restart CMA...)
• K-PCX Sinha et al. (A Population-Based,...)
• L-CMA-ES Auger and Hansen (Performance Evaluation...)
• L-SaDE Qin and Suganthan (Self-Adaptive Differential...)
• SPC-PNX Ballester et al. (Real-Parameter Optimization...)

On the consistency of ranking algorithms by J. Duchi, L. Mackey and M. I. Jordan (admin of MLcomp.org), ICML 2010

Mathematical complexity

1. Open problem
2. ...
3. Polynomial complexity of low degree

L'arbre de la complexite Le Pouvoir Des Mathematiques, La Recherche 2009

Questions for "diagnosis"

Each step is associated with an increasingly important and precise amount of questions in order to localize how far we should go.

From questions to automated analysis

The set of questions form a network of pattern that aims to be used without human supervision.

To organize

• Tricki.org
• cTuning.org
• Optimal Ordered Problem Solver (OOPS) by J. Schmidhuber
• Natural Evolution Strategies by J. Schmidhuber
• related discussion with Raphael on creativity