"... In PAGE 4: ...ttp://www.auml.org/), and self-organised agent communities [55], respectively. Table3... In PAGE 5: ... Four sets of mutants are generated from the seeds by the mutation analysis tool, respectively. The total numbers of mutants generated from each seed are also given in Table3 . In the sequel, we will use Amalthaea suite, Auction suite, Community suite and UNSC suite to refer to their mutant sets, respectively.... ..."

"... In PAGE 4: ... We then trained a Self-Organising Map on the track vectors for these words, using LSA at rank 40, and mapped each word onto its best-matching unit in the trained SOM. The resulting con- figuration of terms is shown in Table2 , and gives an im- pression of the organisation of emotion words in our se- mantic space. This shows some relationship to the tradi- tional arousal-valenceaxes, with valence increasing broadly from left to right and arousal from top to bottom.... ..."

"... In PAGE 5: ... Results for heterogeneous/homogeneous agents These results are specific for one run, although these results can be obtained in many runs. As it can be seen from Table3 heterogeneous agents tend to self-organise taking over each one a role. There is a very interesting aspect to notice: in logged data agents 1 and 4 seem to take a role, but after some time they finally settle in a different role.... ..."

"... In PAGE 5: ...130 enterprise 21 ops 130 mars 21 st 130 observer 21 Table4 : Example SOM Word Clusters The most apparent attribute of these clusters is their small size. Indeed, the self-organising map clusters range from 2 words to a maximum of 9 words in one cluster.... ..."

"... In PAGE 32: ... Table9 -2 Acronyms AI Artificial Intelligence ART Adaptive Resonance Theory: Neural network capable of incremental learning of class prototypes BCS British Computer Society FDA Food and Drug Administration FDI Fault Detection and Identification FFT Fast Fourier Transform GTM Generative Topographic Mapping: self-organised neural network with a noise model GUI Graphical User Interface MLP Multi-layer Perceptron: the most commonly used neural network for static pattern recognition PID Proportional, integral and derivative term: refers to process control SOM Self-organised mapping: neural network used to map high- dimensional data into usually two dimensions, by covering the data with a 2-D flexible surface then projecting the data onto it. SWARM Freeware for optimisation by simulating a large number of interacting software entities, called agents.... ..."

"... In PAGE 32: ...30 DEFINITIONS AND ACRONYMS Table9 -1 Definitions Neuromorphic Computer systems that emulate neuro-physiological circuits Regularise Stabilise data fitting by explicitly penalising model complexity, usually with an additional term in the objective function Sensitivity True positive detection rate Specificity True negative detection rate ROC Receiver Operating Characteristic, a formalism to analyse detection rates across all values of the detection threshold Self-organised model Data density model commonly used for clustering and visualisation Supervised model Empirical model that attempts to replicate decision labels Table 9-2 Acronyms ... ..."

"... In PAGE 6: ...ig. 2. test case isosurfaces The ESFE results shown for EPSOC and the GA are the generation counts at which the minimum result was obtained, not the count for termination, which was fixed. It may be noted from Table1 that EPSOC effectively found the global minimum in 6 out of 7 cases (counting the result on the Crack 2 test case as successful, since it was within 0.03% of the global minimum value).... In PAGE 7: ...62 15 -26.98 5 8e-4 33 Table1 . Best objective functions values obtained in 10 runs, and time taken Conclusions In this paper we have described a simple new Evolutionary Programming algorithm that utilizes concepts of Self-Organised Criticality.... ..."

"... In PAGE 2: ...ffinity for a range of target proteins in E. coli. METHODS Bacterial strains, plasmids and growth conditions. The bacter- ial strains used in this study are described in Table1 . E.... ..."

"... In PAGE 7: ...igure 1. Plots of scores, 8, for highest scoring sequence as a function of window position. The parameters are word length k = 6, window width W = 9, and number of mismatches = 2 (no insertions or deletions permitted). (a) The sequences are aligned on the transcription start site, as shown in Table1 ; (b) the sequences are aligned on the - 10 peak; (c) on the -35 peak, as described in the text. The expected value for random sequences of identical composition is ... ..."