Forgetting Exceptions is Harmful in Language Learning

We show that in language learning, contrary to received wisdom, keeping exceptional training instances in memory can be beneficial for generalization accuracy. We investigate this phenomenon empirically on a selection of benchmark natural language processing tasks: grapheme-to-phoneme conversion, part-of-speech tagging, prepositional-phrase attachment, and base noun phrase chunking. In a first series of experiments we combine memory-based learning with training set editing techniques, in which instances are edited based on their typicality and class prediction strength. Results show that editing exceptional instances (with low typicality or low class prediction strength) tends to harm generalization accuracy. In a second series of experiments we compare memory-based learning and decision-tree learning methods on the same selection of tasks, and find that decision-tree learning often performs worse than memory-based learning. Moreover, the decrease in performance can be linked to the degree of abstraction from exceptions (i.e., pruning or eagerness). We provide explanations for both results in terms of the properties of the natural language processing tasks and the learning algorithms.     

Neural navigation interfaces for Information Retrieval: Are they more than an appealing idea?

Neural networks have recently been proposed for the construction of navigation interfaces for Information Retrieval systems. In this paper, we give an overview of some current research in this area. Most of the cited approaches use (variants) of the well-known Kohonen network. The Kohonen network implements a topology-preserving dimensionality-reducing mapping, which can be applied for information visualization. We identify a number of problems in the application of Kohonen networks for Information Retrieval, most notably scalability, reliability and retrieval effectiveness. To solve these problems we propose to use the Growing Cell Structures network, a variant of the Kohonen network which shows a more flexible adaptation to the domain structure.This network was tested on two standard test-collections, using a combined recall and precision measure, and compared to traditional IR methods such as the Vector Space Model and various clustering algorithms. The network performs at a competitive level of effectiveness, and is suitable for visualization purposes. However, the incremental training procedures for the networks result in a reliability problem, and the approach is computationally intensive. Also, the utility of the resulting maps for navigation will need further improvement.     

Memory-Based Word Sense Disambiguation

We describe a memory-based classification architecture for word sense disambiguation and its application to the SENSEVAL evaluationtask. For each ambiguous word, a semantic word expert isautomatically trained using a memory-based approach. In each expert,selecting the correct sense of a word in a new context is achieved byfinding the closest match to stored examples of this task. Advantagesof the approach include (i) fast development time for word experts,(ii) easy and elegant automatic integration of information sources,(iii) use of all available data for training the experts, and (iv)relatively high accuracy with minimal linguistic engineering.     

Model discrimination for the propionic acid diffusion into hydrogel beads using lifetime confocal laser scanning microscopy

Hydrogels are increasingly used for the entrapment of biocatalysts, especially for the use of enzymes in organic solvent systems. Hence, it is necessary to understand the influence of the hydrogel matrix on the transport kinetics (diffusion and phase transfer) and on the reaction kinetics (activity and stability of the enzymes). Here, the diffusion of propionic acid into Ca-alginate hydrogel beads is studied using a structured methodology, model-based experimental analysis (MEXA).The basis for all further investigations is a rigorous model of all processes occurring during the diffusion of the propionic acid into the Ca-alginate hydrogel bead. As competing model assumptions the Fickian and Nernst-Planck diffusion laws are integrated. Then, an optimal experimental design is performed to determine those experiments that allow an efficient discrimination of the rival model candidates. The change in pH-value over time resulting from the diffusion of propionic acid into the hydrogel bead is observed in the centre of the hydrogel bead. The fluorescent pH-indicator resorufin shows pH-dependent lifetimes that are measured by time-correlated single photon counting using a confocal laser scanning microscope at high temporal resolution (every 1 s). To our knowledge lifetime confocal laser scanning microscopy is used for the first time in this study for the quantification of dynamic pH-changes in macroscopically large particles. The model parameters for each candidate model are estimated based on 16 independent experimental data sets. Finally, the likelihood function is used to evaluate and discriminate the competing diffusion laws.For the diffusion of propionic acid into spherical Ca-alginate hydrogel beads, it could be shown that Fickian diffusion is not able to describe the process accurately in contrast to Nernst-Planck diffusion. The influence of the Ca-alginate hydrogel density was found to be insignificant for the estimated diffusion coefficient.     

Systematic determination of intrinsic reaction parameters in enzyme immobilizates

For the rational design of processes using immobilized enzymes a mechanistic kinetic model is required, which accounts for all kinetic and thermodynamic phenomena, including the enzyme reaction, the mass transfer of the reactants between both phases, and their diffusion inside the immobilizate. For the example of enzymes immobilized in hydrogel beads suspended in an organic solvent, such a mechanistic kinetic model was obtained by a model-based experimental analysis approach. It was proven that the usually applied concentration measurements in the bulk phase are not sufficient to draw mechanistic conclusions. The most suitable measurement technique was found to be the quantification of the concentration along the radius of the hydrogel bead. These line scans, achieved by two-photon laser scanning microscopy, for the first time allowed to estimate intrinsic reaction and mass transfer parameters simultaneously. Thus, the obtained intrinsic parameter estimates for the biphasic hydrogel system could be directly compared with those obtained in individual systems. This comparison revealed for the first time that the enzyme reaction was not significantly affected by the mild hydrogel encapsulation. However, a significant impact on the transport parameters was found that underlines the need for analyzing the real reaction system using mechanistic models.