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.

Economic Valuation of Ecosystem Services: Conflicts in Classification

4 p.

Monitoring and classification of lakes in Finland

In Finland, as in other member countries of the European Union, preparations for implementing the EC Water Framework Directive (WFD) have begun. The article describes the current monitoring and classification strategies for Finnish Lakes.

A preliminary classification of lake types in Northern Ireland

The EC Water Framework Directive (WFD) introduces the concept of the ecological status of surface waters. The WFD requires that water bodies within an ecoregion are divided into waterbody "types" or "ecotypes". The waterbody type is determined by physico-chemical descriptors. A group of representatives from two government environmental agencies (Environment and Heritage Service and the Industrial Research and Technology Unit) and the two universities (Queen's University Belfast and the University of Ulster) in Northern Ireland, has been established to develop methods for measuring the ecological status of lakes, for the purposes of the WFD. The work here is that contributing to the first objective classification into waterbody types.

On the classification of lakes and lake-like bodies of the Ukraine [Translation from: Gidrobiologicheskie Zhurnal, Kiev 19(2) 100-101, 1983]

In the Ukraine there are several thousand large, medium and small lakes and lake-like reservoirs, distinguished by origin, salinity, regional position, productivity and by construction a significant number of large and small water bodies, ponds and industrial reservoirs of variable designation. The problem of national systems necessitates the creation of specific schemes and classifications. Classifying into specific types of reservoir by means of suitable specifications is required for planning national measures with the objective of the rational utilisation of natural resources. It is now necessary to consider the present-day characteristics of Ukranian lakes. In the case of the Ukraine it is possible to use two approaches - genetical and ecological. This paper uses the genetical system to classify the lake-like water bodies of the Ukraine.

The classification and prediction of macroinvertebrate communities in British rivers

This article describes the progress of the River Communities Project which commenced in 1977. This project aimed to develop a sensitive and practical system for river site classification using macroinvertebrates as an objective means of appraising the status of British rivers. The relationship between physical and chemical features of sites and their biological communities were examined. Sampling was undertaken on 41 British rivers. Ordination techniques were used to analyze data and the sites were classified into 16 groups using multiple discrimination analysis. The potential for using the environmental data to predict to which group a site belonged and the fauna likely to be present was investigated.

On the optimal usage of labelled examples in semi-supervised multi-class classification problems

In recent years, the performance of semi-supervised learning has been theoretically investigated. However, most of this theoretical development has focussed on binary classification problems. In this paper, we take it a step further by extending the work of Castelli and Cover [1] [2] to the multi-class paradigm. Particularly, we consider the key problem in semi-supervised learning of classifying an unseen instance x into one of K different classes, using a training dataset sampled from a mixture density distribution and composed of l labelled records and u unlabelled examples. Even under the assumption of identifiability of the mixture and having infinite unlabelled examples, labelled records are needed to determine the K decision regions. Therefore, in this paper, we first investigate the minimum number of labelled examples needed to accomplish that task. Then, we propose an optimal multi-class learning algorithm which is a generalisation of the optimal procedure proposed in the literature for binary problems. Finally, we make use of this generalisation to study the probability of error when the binary class constraint is relaxed.