The improved management of small-scale cage culture in Asia: final technical report

The purpose of the project is to develop sustained small-scale cage fish culture in inland and coastal waters through improved understanding of the social, institutional and resource environment of resource poor groups. Two Asian countries, Bangladesh (inland systems) and Vietnam (marine), were studied with this workshop bringing together both sides of the project together with representatives of collaborative institutions, government departments and universities. Addressing the overall aim of producing guidelines for the planning and extension of cage aquaculture in Asia a combination of group work and plenary discussion was conducted producing the following outputs. 1) An assessment of cage aquaculture potential, 2) Development options for small-scale cage culture, 3) A review of tools and methodologies and 4) Policy initiatives for sustainable cage culture development. Key issues raised were the use of outputs as a guide to be adapted to regional circumstances to facilitate farmer and extension worker discussion and not as a rigid methodology. The degree of linkage between development, research and government institutions was also considered a crucial factor in benefiting the research and development of cage culture at the local, regional and national level and vital in affecting the future policies by both development and government institutions. [PDF contains 242 pages]

Estimating abundances of 0-group western Baltic cod by using pound net fisheries

Nearshore 0-group western Baltic cod are frequently caught as bycatch in the commercial pound net fishery. Pound net fishermen from the Danish Isle of Funen and Lolland and the German Isle of Fehmarn have recorded their catches of small cod between September and December 2008. Abundance patterns were analysed, particularly concerning the influence of abiotic factors (hydrography, meteorology) and the differences between sampling sites. Catch per unit effort (CPUE) differed by site and location, whereas CPUE were highest at Lolland. Correlation between catch and wind/currents were generally weak. However, wind directions and current speeds seem to affect the catch rates. Finally an algorithm was developed to calculate a recruitment index for western Baltic cod recruitment success based on previous analyses.

Group size and decision rules in legislative bargaining

We conduct experiments to investigate the effects of different majority requirements on bargaining outcomes in small and large groups. In particular, we use a Baron-Ferejohn protocol and investigate the effects of decision rules on delay (number of bargaining rounds needed to reach agreement) and measures of "fairness" (inclusiveness of coalitions, equality of the distribution within a coalition). We find that larger groups and unanimity rule are associated with significantly larger decision making costs in the sense that first round proposals more often fail, leading to more costly delay. The higher rate of failure under unanimity rule and in large groups is a combination of three facts: (1) in these conditions, a larger number of individuals must agree, (2) an important fraction of individuals reject offers below the equal share, and (3) proposers demand more (relative to the equal share) in large groups.

Space-frequency coupling, conical waves, and small-scale filamentation in water

Numerical simulations of fs laser propagation in water have been made to explain the small-scale filaments in water we have observed by a nonlinear fluorescence technique. Some analytical descriptions combined with numerical simulations show that a space-frequency coupling mainly from the interplay among self-phase modulation, dispersion and phase mismatching will reshape the laser beam into a conical wave which plays a major role of energy redistribution and can prevent laser beam from self-guiding over a long distance. An effective group velocity dispersion is introduced to explain the pulse broadening and compression in the filamentation. (c) 2005 American Institute of Physics.

Learning and using taxonomies for visual and olfactory classification

Humans are able of distinguishing more than 5000 visual categories even in complex environments using a variety of different visual systems all working in tandem. We seem to be capable of distinguishing thousands of different odors as well. In the machine learning community, many commonly used multi-class classifiers do not scale well to such large numbers of categories. This thesis demonstrates a method of automatically creating application-specific taxonomies to aid in scaling classification algorithms to more than 100 cate- gories using both visual and olfactory data. The visual data consists of images collected online and pollen slides scanned under a microscope. The olfactory data was acquired by constructing a small portable sniffing apparatus which draws air over 10 carbon black polymer composite sensors. We investigate performance when classifying 256 visual categories, 8 or more species of pollen and 130 olfactory categories sampled from common household items and a standardized scratch-and-sniff test. Taxonomies are employed in a divide-and-conquer classification framework which improves classification time while allowing the end user to trade performance for specificity as needed. Before classification can even take place, the pollen counter and electronic nose must filter out a high volume of background “clutter” to detect the categories of interest. In the case of pollen this is done with an efficient cascade of classifiers that rule out most non-pollen before invoking slower multi-class classifiers. In the case of the electronic nose, much of the extraneous noise encountered in outdoor environments can be filtered using a sniffing strategy which preferentially samples the visensor response at frequencies that are relatively immune to background contributions from ambient water vapor. This combination of efficient background rejection with scalable classification algorithms is tested in detail for three separate projects: 1) the Caltech-256 Image Dataset, 2) the Caltech Automated Pollen Identification and Counting System (CAPICS) and 3) a portable electronic nose specially constructed for outdoor use.

Convex analysis for minimizing and learning submodular set functions

The connections between convexity and submodularity are explored, for purposes of minimizing and learning submodular set functions.

First, we develop a novel method for minimizing a particular class of submodular functions, which can be expressed as a sum of concave functions composed with modular functions. The basic algorithm uses an accelerated first order method applied to a smoothed version of its convex extension. The smoothing algorithm is particularly novel as it allows us to treat general concave potentials without needing to construct a piecewise linear approximation as with graph-based techniques.

Second, we derive the general conditions under which it is possible to find a minimizer of a submodular function via a convex problem. This provides a framework for developing submodular minimization algorithms. The framework is then used to develop several algorithms that can be run in a distributed fashion. This is particularly useful for applications where the submodular objective function consists of a sum of many terms, each term dependent on a small part of a large data set.

Lastly, we approach the problem of learning set functions from an unorthodox perspective---sparse reconstruction. We demonstrate an explicit connection between the problem of learning set functions from random evaluations and that of sparse signals. Based on the observation that the Fourier transform for set functions satisfies exactly the conditions needed for sparse reconstruction algorithms to work, we examine some different function classes under which uniform reconstruction is possible.

The Green Paradox and Learning-by-doing in the Renewable Energy Sector

21 p.