This is a wasted opportunity. The basic concept -- integration of Neural Networks into nuclear and high energy experimental physics) -- is sound, interesting and appropriate for the ITR program. Further their architecture with a library of filters and a browser based toolkit is good.

Unfortunately there is essentially no discussion of the key IT issues -- structure of classification algorithms or design of the portal. Rather the proposal is largely a catalog of opportunities (this or that physics reaction) for applying neural network filters. Further the proposal boldly (naively?) states (page 9) that year 1 will be devoted to senior researchers and students learning the field. This is consistent with the lack of understanding of IT shown in the proposal but not in my humble opinion "allowed" in a winning proposal.
	We can note that
1) There is no documented expertise in Neural Networks on the team.
2) There is no documented understanding that neural networks are just one
"advanced" classification system capable of developing such filters. In
the last 15 years or so of research in neural network classifers, much
sophicated work has been done and a broad understanding of best practice
should be documented in the proposal.
3) There is no documented interdisciplinary collaboration with the IT
research field.

	I am a little surprised by the proposal weaknesses as the proposers 
appear to be involved with succesful application of the basic neural
network ideas at Brookhaven. Perhaps this is connected with the departure
of key people alluded to in proposal. One could give a more detailed
critique but I think the fate of this proposal is clear -- it should be
rejected. I would encourage the field to develop the necessary links with
IT community and resubmit a proposal with the physics excellence of this
team strengthened by IT expertise.