This is a step towards an ongoing effort to Clean My Desk. And also to cut back on the frivolous increase in the world’s heat I generate by indiscriminately picking books up at the library and hauling them back here (only to be told they’re overdue without being read). I’m not reviewing; rather, noting interesting articles in the books.

I’ve already had some words, months back, about the ridiculous prices charged by certain Northern European Publishing Clans, and I don’t want you to consider buying these books for an instant. The links to the left are offered more in the spirit of a public shame through literal fact, not an approbation. You could go to Amazon and buy some other crap—some useful crap, if you want. I wouldn’t mind that. Because Amazon would pay me, so I would be less minding that, for the paying. I am not at all about minding paying.

But in general these are books that are for the bibliographic padding of the contributors, not to be read, not even to be referenced physically, but merely to be socked away on some shelf in a basement High Density Storage, and to see the light of day only when some idiot (like me) gets a dose of undirected curiosity.

So: Expensive academic press vanity doorstops = dumb. But: Authors in said books = sometimes very interesting. To kill π birds with one stone, I’ll call out some of the articles and chapters and equations that catch my eye, and briefly discuss them. And offer links to free preprints online, as available.

I’m starting with the fattest. Clear that space off quick.

[I'll add the telegraphic notices as I have time today. These books are due, after all.]

AI 2004: Advances in Artificial Intelligence is one of those catch-all proceedings volumes that is full of this and that. As local Specialist in This and That, I like it. Not all. Here are some contributions that at least caught my eye:

• “Critical damage reporting in intelligent sensor networks” by Jiaming Li, Ying Guo, and Geoff Poulton. [not available online!?] Wrap a spacecraft in a “skin” of locally-connected sensor agents. When a little meteor or a wayward space bolt strikes it, they yell at each other. How do you arrange them so that the collective network structure can understand (and communicate) the difference between random failure, minor damage and critical damage? Especially when you don’t know where the damage will be, and if it will affect the crucial “portal” communicator agents. You evolve a pheremone-directed signaling route on the fly.
• “Combining Bayesian networks, k nearest neighbours algorithm and attribute selection for gene expression data analysis”, B. Sierra, E. Lazkano, J. M. Martínez-Otzeta, and A. Astigarraga. [also not online!? sheesh.] Biology used to be so simple, so elegant, so observational. Now it’s burdened with data lacking knowledge, and all those years of complaint that “Math is hard; let’s do biology!” have wrought a fearsome slack, being taken up by folks in other disciplines. Like these. The problem here: Gene expression chips (AffyMetrix and others) result in thousands of data points for every experiment. Each of those 2000+ numbers is (arguably) the expression level of a certain RNA species in vivo. How do you take a 2000-dimensional timeseries, and reconstruct a genetic regulatory network from it? The authors’ response (roughly) is an iterative variable-selection and learning cycle: identify a small set of salient (influential explanatory) genes from the mess, and add them to a database; build naive Bayes models of the databased gene dynamics using the complete dataset, in order to identify new genes to add to the mix. Iterate.