Reducing performance bias for unbalanced text mining

In text categorization applications, class imbalance, which refers to an uneven data distribution where one class is represented by far more less instances than the others, is a commonly encountered problem. In such a situation, conventional classifiers tend to have a strong performance bias, which results in high accuracy rate on the majority class but very low rate on the minorities. An extreme strategy for unbalanced, learning is to discard the majority instances and apply one-class classification to the minority class. However, this could easily cause another type of bias, which increases the accuracy rate on minorities by sacrificing the majorities. This paper aims to investigate approaches that reduce these two types of performance bias and improve the reliability of discovered classification rules. Experimental results show that the inexact field learning method and parameter optimized one-class classifiers achieve more balanced performance than the standard approaches.

Minimization of reversible adder circuits

Losing information causes losing power. Information is lost when the input vector cannot be uniquely recovered from the output vector of a combinational circuit. The input vector of reversible circuit can be uniquely recovered from the output vector. In this study we have emphasized on the design of reversible adder circuits that is efficient in terms of gate count, garbage outputs and quantum cost and that can be technologically mapped. It has been analyzed and demonstrated that the results of our proposed adder circuits shows better performance compared to similar type of existing designs. Technology independent equations required to evaluate these circuits have also been given.

Reducing performance bias by intrinsically insensitive learning for unbalanced text mining

This thesis proposes three effective strategies to solve the significant performance-bias problem in imbalance text mining: (1) creation of a novel inexact field learning algorithm to overcome the dual-imbalance problem; (2) introduction of the one-class classification-framework to optimize classifier-parameters, and (3) proposal of a maximal-frequent-item-set discovery approach to achieve higher accuracy and efficiency.

Improving reliability of unbalanced text mining by reducing performance bias

Class imbalance in textual data is one important factor that affects the reliability of text mining. For imbalanced textual data, conventional classifiers tend to have a strong performance bias, which results in high accuracy rate on the majority class but very low rate on the minorities. An extreme strategy for unbalanced learning is to discard the majority instances and apply one-class classification to the minority class. However, this could easily cause another type of bias, which increases the accuracy rate on minorities by sacrificing the majorities. This chapter aims to investigate approaches that reduce these two types of performance bias and improve the reliability of discovered classification rules. Experimental results show that the inexact field learning method and parameter optimized one class classifiers achieve more balanced performance than the standard approaches.

The microbiological and sustainability effects of washing anaesthesia breathing circuits less frequently

In the presence of single-use airway filters, we quantified anaesthetic circuit aerobic microbial contamination rates when changed every 24 h, 48 h and 7 days. Microbiological samples were taken from the interior of 305 anaesthetic breathing circuits over a 15-month period (3197 operations). There was no significant difference in the proportion of contaminated circuits when changed every 24 h (57/105 (54%, 95% CI 45–64%)) compared with 48 h (43/100 (43%, 95% CI 33–53%, p = 0.12)) and up to 7 days (46/100 (46%, 95% CI 36–56%, p = 0.26)). Median bacterial counts were not increased at 48 h or 7 days provided circuits were routinely emptied of condensate. Annual savings for one hospital (six operating theatres) were $AU 5219 (£3079, €3654, $US 4846) and a 57% decrease in anaesthesia circuit steriliser loads associated with a yearly saving of 2760 kWh of electricity and 48 000 l of water. Our findings suggest that extended circuit use from 24 h up to 7 days does not significantly increase bacterial contamination, and is associated with labour, energy, water and financial savings.

Rectenna circuits for RF energy harvesting in miniature DBS devices.

 Development of an optimum rectenna for radio frequency energy harvesting in miniature head-mountable deep brain stimulation (DBS) devices. The designed miniature rectenna can operate a DBS device without battery for murine preclinical research. The battery-less operation of the device eliminates battery related difficulties.

Fabrication of force sensor circuits on wearable conductive textiles

This paper discusses design and fabrication processes in the development of a wearable and flexible conductive resistive sensor. The design and development of the sensor involve the use of Sn-Ag-Cu (SAC)plated Nylon fabric, precisionfused deposition modeling(FDM) using silicone and petrolatum for etch-resistant masks using the EnvisionTEC GmbH Bioplotter, and wet etching using Chromium, Ammonium Persulphate, and Salt-Vinegar etching solutions. Preliminary testing with other mask types, development processes, and sensor design approaches for various applications are discussed.

Distributed generation capacity planning for distribution networks to minimize energy loss: an unbalanced multi-phase optimal power flow based approach

This chapter presents an unbalanced multi-phase optimal power flow (UMOPF) based planning approach to determine the optimum capacities of multiple distributed generation units in a distribution network. An adaptive weight particle swarm optimization algorithm is used to find the global optimum solution. To increase the efficiency of the proposed scheme, a co-simulation platform is developed. Since the proposed method is mainly based on the cost optimization, variations in loads and uncertainties within DG units are also taken into account to perform the analysis. An IEEE 123 node distribution system is used as a test distribution network which is unbalanced and multi-phase in nature, for the validation of the proposed scheme. The superiority of the proposed method is investigated through the comparisons of the results obtained that of a Genetic Algorithm based OPF method. This analysis also shows that the DG capacity planning considering annual load and generation uncertainties outperform the traditional well practised peak-load planning.