An empirical study of integration activities in distributions of open source software

Reuse of software components, either closed or open source, is considered to be one of the most important best practices in software engineering, since it reduces development cost and improves software quality. However, since reused components are (by definition) generic, they need to be customized and integrated into a specific system before they can be useful. Since this integration is system-specific, the integration effort is non-negligible and increases maintenance costs, especially if more than one component needs to be integrated. This paper performs an empirical study of multi-component integration in the context of three successful open source distributions (Debian, Ubuntu and FreeBSD). Such distributions integrate thousands of open source components with an operating system kernel to deliver a coherent software product to millions of users worldwide. We empirically identified seven major integration activities performed by the maintainers of these distributions, documented how these activities are being performed by the maintainers, then evaluated and refined the identified activities with input from six maintainers of the three studied distributions. The documented activities provide a common vocabulary for component integration in open source distributions and outline a roadmap for future research on software integration.     

Bed morphology, flow structure, and sediment transport at the outlet of Lake Huron and in the upper St. Clair River

An integrated multibeam echo sounder and acoustic Doppler current profiler field survey was conducted in July 2008 to investigate the morphodynamics of the St. Clair River at the outlet of Lake Huron. The principal morphological features of the upper St. Clair River included flow-transverse bedforms that appear weakly mobile, erosive bedforms in cohesive muds, thin non-cohesive veneers of weakly mobile sediment that cover an underlying cohesive (till or glacio-lacustrine) surface, and vegetation that covers the bed. The flow was characterized by acceleration as the banks constrict from Lake Huron into the St. Clair River, an approximately 1500-m long region of flow separation downstream from the Blue Water Bridge, and secondary flow connected to: i) channel curvature; ii) forcing of the flow by local bed topography, and iii) flow wakes in the lee side of ship wrecks. Nearshore, sand-sized, sediment from Lake Huron was capable of being transported into, and principally along, the banks of the upper St. Clair River by the measured flow. A comparison of bathymetric surveys conducted in 2007 and 2008 identifies that the gravel bed does undergo slow downstream movement, but that this movement does not appear to be generated by the mean flow, and could possibly be caused by ship-propeller-induced turbulence. The study results suggest that the measured mean flow and dredging within the channel have not produced major scour of the upper St. Clair River and that the recent fall in the level of Lake Huron is unlikely to have been caused by these mechanisms.     

Inductive transfer with context-sensitive neural networks

Context-sensitive Multiple Task Learning, or csMTL, is presented as a method of inductive transfer which uses a single output neural network and additional contextual inputs for learning multiple tasks. Motivated by problems with the application of MTL networks to machine lifelong learning systems, csMTL encoding of multiple task examples was developed and found to improve predictive performance. As evidence, the csMTL method is tested on seven task domains and shown to produce hypotheses for primary tasks that are often better than standard MTL hypotheses when learning in the presence of related and unrelated tasks. We argue that the reason for this performance improvement is a reduction in the number of effective free parameters in the csMTL network brought about by the shared output node and weight update constraints due to the context inputs. An examination of IDT and SVM models developed from csMTL encoded data provides initial evidence that this improvement is not shared across all machine learning models.     

A novel fluorescence lifetime imaging system that optimizes photon efficiency

Fluorescence lifetime imaging (FLIM) is a powerful microscopy technique for providing contrast of biological and other systems by differences in molecular species or their environments. However, the cost of equipment and the complexity of data analysis have limited the application of FLIM. We present a mathematical model and physical implementation for a low cost digital frequency domain FLIM (DFD-FLIM) system, which can provide lifetime resolution with quality comparable to time-correlated single photon counting methods. Our implementation provides data natively in the form of phasors. On the basis of the mathematical model, we present an error analysis that shows the precise parameters for maximizing the quality of lifetime acquisition, as well as data to support this conclusion. The hardware and software of the proposed DFD-FLIM method simplifies the process of data acquisition for FLIM, presents a new interface for data display and interpretation, and optimizes the accuracy of lifetime determination.     

Electrochemical studies of octadecanethiol and octanethiol films on variable surface area mercury sessile drops

Octadecanethiol (ODT) and octanethiol (OT) films at the mercury-electrolyte interface are examined using cyclic voltammetry and differential capacitance measurements at a single frequency. A mercury flow-system is used to alter the volume, and therefore, the surface area and surface pressure of the mercury electrode. Manipulation of the mercury electrode's volume enables the introduction and removal of defects in the insulating thiol films. OT and ODT film behavior are contrasted under conditions of expansion and contraction. ODT forms extremely impermeable layers that allow 1000 time less redox probe current than seen on uncoated drops. Expansion of the mercury electrode to increase the electrode surface area produces defects and pinholes in the thiol film. These defects are almost completely removed when the drop is compressed back to its initial surface area. OT also forms insulating films on mercury sessile drops, however these films contain more defects than ODT films. While expansion of an OT-coated mercury drop increases redox probe current, recompression of the drop does not return the film to its initial condition. Pinholes and defects in the OT and ODT films can also be produced by cycling to negative potentials, which produce abrupt stripping peaks.     

REVISE: A Tool for Measuring and Mitigating Bias in Visual Datasets

International Journal of Computer Vision - Machine learning models are known to perpetuate and even amplify the biases present in the data. However, these data biases frequently do not become...