Improving class probability estimates for imbalanced data

Obtaining good probability estimates is imperative for many applications. The increased uncertainty and typically asymmetric costs surrounding rare events increase this need. Experts (and classification systems) often rely on probabilities to inform decisions. However, we demonstrate that class probability estimates obtained via supervised learning in imbalanced scenarios systematically underestimate the probabilities for minority class instances, despite ostensibly good overall calibration. To our knowledge, this problem has not previously been explored. We propose a new metric, the stratified Brier score, to capture class-specific calibration, analogous to the per-class metrics widely used to assess the discriminative performance of classifiers in imbalanced scenarios. We propose a simple, effective method to mitigate the bias of probability estimates for imbalanced data that bags estimators independently calibrated over balanced bootstrap samples. This approach drastically improves performance on the minority instances without greatly affecting overall calibration. We extend our previous work in this direction by providing ample additional empirical evidence for the utility of this strategy, using both support vector machines and boosted decision trees as base learners. Finally, we show that additional uncertainty can be exploited via a Bayesian approach by considering posterior distributions over bagged probability estimates.     

Styryl Dyes for Synthetic-polymer Fibres.

The synthesis of a series of dyes derived from reaction of 2-methyl-4-(N-ethyl-N-hydroxyethyl)amino, lidicyano-styrene with acid chlorides, chloroformates, isocyanates and isothiocyanates is described. Data relevant to dye characterization by electron-impact induced fragmentation is reported. Dyeing and fastness properties of the dyes on cellulose secondary acetate, cellulose triacetate, nylon 6.6 and polyester fibres are discussed in terms of dye structure. Comparison is made between some related dyes derived from 2-methyl-4-(N-ethyl-N-β-aminoethyljamino-β,βdicyanostyrene.     

The electret effect in cellulose acetate reverse osmosis membranes

The electret potentials developed by reverse osmosis electret membranes help control the undesirable deposition of charged colloidal particles on the membrane surfaces during membrane desalination. These antifouling electret membranes should help prevent the costly flux declines normally associated with deposition of colloidal iron oxides on the reverse osmosis membrane surfaces. Homocharge and heterocharge behavior of cellulose acetate membrane electrets have been studied. Asymmetric reverse osmosis membranes and dense membrane films were studied. The homocharge and heterocharge of cellulose acetate reverse osmosis electret membranes have been explained.     

Chemical inhibitors when timing is critical: a pharmacological concept for the maturation of T cell contacts

Cellular signal transduction proceeds through a complex network of molecular interactions and enzymatic activities. The timing of these molecular events is critical for the propagation of a signal and the generation of a specific cellular response. To define the timing of signalling events, we introduce the combination of high-resolution confocal microscopy with the application of small-molecule inhibitors at various stages of signal transduction in T cells. Inhibitors of Src-family tyrosine kinases and actin dynamics were employed to dissect the role of the lymphocyte-specific tyrosine kinase Lck in the formation and maintenance of T cell receptor/CD3-dependent contacts. Anti-CD3epsilon-coated coverslips served as a highly defined stimulus. The kinetics of the recruitment of the yellow fluorescent protein-tagged signalling protein ZAP-70 were detected by high-resolution confocal microscopy. The analysis revealed that at 5 min after receptor engagement, Lck activity was required for maintenance of contacts. In contrast, after 20 min of receptor engagement, the contacts were Lck-independent. The relevance of the timing of inhibitor application provides a pharmacological concept for the maturation of T cell-substrate contacts.     

Lightweight dynamic symmetry breaking

Symmetries in constraint problems present an opportunity for reducing search. This paper presents Lightweight Dynamic Symmetry Breaking, an automatic symmetry breaking method that is efficient enough to be used as a default, since it never yields a major slowdown while often giving major performance improvements. This is achieved by automatically exploiting certain kinds of symmetry that are common, can be compactly represented, easily and efficiently processed, automatically detected, and lead to large reductions in search. Moreover, the method is easy to implement and integrate in any constraint system. Experimental results show the method is competitive with the best symmetry breaking methods without risking poor performance.     

Copolyester studies. I. Preparation and characterization of tetramethylene terephthalate–tetramethylene sebacate copolymers

Tetramethylene terephthalate–tetramethylene sebacate copolymers containing up to 20 mol % sebacate have been prepared and characterized. Molecular weights and distributions have been evaluated using viscometry and gel-permeation chromatography. Compositions have been established by NMR spectroscopy. Thermal properties have been studied by differential scanning calorimetry and dynamic mechanical methods. Melting and glass-transition temperatures and moduli are discussed in terms of the structural differences, particularly the effect of composition on chain flexibility. Density and thermal methods of crystallinity determination are critically discussed for these systems.     

The behaviour of PbO2 in concentrated sulphuric acid

Linear sweep experiments on Pb in H₂SO₄ at concentrations in excess of 5 mol dm^–3 have been conducted using computer controlled techniques. Measurements have indicated that the maximum charge in the PbO₂ reduction peak occurs at 5 mol dm^–3, the available capacity decreasing with the concentration of H₂SO₄.     

PHASE-SPACE DIFFUSION EQUATIONS FOR SINGLE BROWNIAN PARTICLE MOTION NEAR SURFACES

In numerous physical processes involving the motion of micron and submicron sized particles near surfaces, such as the filtration of hydrosols and aerosols, the particle motion is the net result of the combined effects of fluid convection, external forces, particle inertia, Brownian particle motion, and particle-surface fluid dynamic interactions. The most general method of describing particle motion under the combined action of these effects is through the so-called Fokker-Planck equation. In the absence of particle-surface fluid dynamic interactions, the Fokker-Planck equation is well-known, and it has been applied in a general way to problems involving the adsorption or deposition of Brownian particles onto surfaces through a solution technique known as the Brownian dynamics simulation method.

In this study, the Fokker-Planck equation for Brownian particle motion near surfaces is generalized to include particle-surface fluid dynamic interactions. The Fokker-Planck equation is shown to follow from the Liouville equation for the Brownian particle and n-fluid molecules present in the system, thus, establishing a firm theoretical foundation for the Fokker-Planck equation and the various other phase-space diffusion equations that follow from it.

Based on diagonalization of the Fokker-Planck equation, its short-time behavior is also derived here which enables a generalization of the Brownian dynamics method for the study of particle motion near surfaces including fluid dynamic interactions. Additionally, a perturbation solution of the Fokker-Planck equation under the conditions of small, but finite particle Stokes number is also derived. These solutions are shown to agree with previously given representations of the Smoluchowski or convective-diffusion equation for Brownian particle motion near surfaces, as well as with inertial corrections to the Smoluchowski equation available in the literature. This latter equation is also generalized here to include particle-surface fluid dynamic interactions.     

Adsorption of Cationic Dyes by Acrylic Films II—Kinetics of Dyeing

Concentration distributions of basic dyes in a film of acrylic polymer have been determined using a microdensitometric technique. The diffusion coefficients calculated from them are dependent on concentration. An explanation of the dependence is offered based on the ionic transfer of hydrogen (or sodium) ions for dye. From the concentration dependence an average diffusion coefficient was calculated and found to agree with that calculated from rates of dyeing. Rates of dyeing of a binary mixture have demonstrated the competitive effect predicted by the diffusion equation derived for the individual dyes.