Global optimization of feature weights and the number of neighbors that combine in a case-based reasoning system

Abstract: Case-based reasoning (CBR) often shows significant promise for improving the effectiveness of complex and unstructured decision-making. Consequently, it has been applied to various problem-solving areas including manufacturing, finance and marketing. However, the design of appropriate case indexing and retrieval mechanisms to improve the performance of CBR is still a challenging issue. Most previous studies on improving the effectiveness of CBR have focused on the similarity function aspect or optimization of case features and their weights. However, according to some of the prior research, finding the optimal k parameter for the k-nearest neighbor is also crucial for improving the performance of the CBR system. Nonetheless, there have been few attempts to optimize the number of neighbors, especially using artificial intelligence techniques. In this study, we introduce a genetic algorithm to optimize the number of neighbors that combine, as well as the weight of each feature. The new model is applied to the real-world case of a major telecommunication company in Korea in order to build a prediction model for customer profitability level. Experimental results show that our genetic-algorithm-optimized CBR approach outperforms other artificial intelligence techniques for this multi-class classification problem.     

Reduction of aflatoxins by Korean soybean paste and its effect on cytotoxicity and reproductive toxicity--part 1. Inhibition of growth and aflatoxin production of Aspergillus parasiticus by Korean soybean paste (Doen-jang) and identification of the active component

The inhibitory effect of methanol extract of Korean soybean paste on the mold growth and aflatoxin production of a toxigenic strain of Aspergillus parasiticus ATCC 15517 was studied using different concentrations of the extract in yeast-extract sucrose broth. While inhibition in mold growth due to increasing the concentration of the extract was observed, the more remarkable effect was the inhibition of aflatoxin production. Reduction of mycelial weight as a result of addition of the extract was observed to range between 1.5 to 12.9% while reduction of aflatoxin production quantified by high-performance liquid chromatography ranged from 14.3 to 41.7%. Five percent of the extract significantly reduced aflatoxin production at the end of the incubation period (P < 0.05), although the effect on mycelial growth was less pronounced. This study indicates that soybean paste could also be an effective inhibitor of aflatoxin production even though mycelial growth may be permitted. The main active component identified by gas chromatography-mass spectroscopy was linoleic acid.     

Microfluidic Synthesis of pH‐Sensitive Multicompartmental Microparticles for Multimodulated Drug Release

Stimuli‐responsive carriers releasing multiple drugs have been researched for synergistic combinatorial cancer treatment with reduced side‐effects. However, previously used drug carriers have limitations in encapsulating multiple drug components in a single carrier and releasing each drug independently. In this work, pH‐sensitive, multimodulated, anisotropic drug carrier particles are synthesized using an acid‐cleavable polymer and stop‐flow lithography. The particles exhibit a faster drug release rate at the acidic pH of tumors than at physiological pH, demonstrating their potential for tumor‐selective drug release. The drug release rate of the particles can be adjusted by controlling the monomer composition. To accomplish multimodulated drug release, multicompartmental particles are synthesized. The drug release profile of each compartment is programmed by tailoring the monomer composition. These pH‐sensitive, multicompartmental particles are promising drug carriers enabling tumor‐selective and multimodulated release of multiple drugs for synergistic combination cancer therapy.     

A Fully Integrated Thin‐Film Inductor and Its Application to a DC‐DC Converter

This paper presents a simple process to integrate thin‐film inductors with a bottom NiFe magnetic core. NiFe thin films with a thickness of 2 to 3 μm were deposited by sputtering. A polyimide buffer layer and shadow mask were used to relax the stress of the NiFe films. The fabricated double spiral thin‐film inductor showed an inductance of 0.49 μH and a Q factor of 4.8 at 8 MHz. The DC‐DC converter with the monolithically integrated thin‐film inductor showed comparable performances to those with sandwiched magnetic layers. We simplified the integration process by eliminating the planarization process for the top magnetic core. The efficiency of the DC‐DC converter with the monolithic thin‐film inductor was 72% when the input voltage and output voltage were 3.5 V and 6 V, respectively, at an operating frequency of 8 MHz.     

Race Classification Using Deep Learning

Race classification is a long-standing challenge in the field of face image analysis. The investigation of salient facial features is an important task to avoid processing all face parts. Face segmentation strongly benefits several face analysis tasks, including ethnicity and race classification. We propose a race-classification algorithm using a prior face segmentation framework. A deep convolutional neural network (DCNN) was used to construct a face segmentation model. For training the DCNN, we label face images according to seven different classes, that is, nose, skin, hair, eyes, brows, back, and mouth. The DCNN model developed in the first phase was used to create segmentation results. The probabilistic classification method is used, and probability maps (PMs) are created for each semantic class. We investigated five salient facial features from among seven that help in race classification. Features are extracted from the PMs of five classes, and a new model is trained based on the DCNN. We assessed the performance of the proposed race classification method on four standard face datasets, reporting superior results compared with previous studies.     

Novel Diffusion‐Barrier Materials Against Oxygen for High‐Density Dynamic Random Access Memory Capacitors

A new design concept for diffusion barriers in high‐density memory capacitors is suggested, and both RuTiN (RTN) and RuTiO (RTO) films are proposed as sacrificial oxygen diffusion barriers. The newly developed RTN and RTO barriers show a much lower sheet resistance than various other barriers, including binary and ternary nitrides (reported by others), up to 800 °C, without a large increase in the resistance. For both the Pt/RTN/TiSi_x/n⁺⁺poly‐plug/n⁺ channel layer/Si and the Pt/RTO/RTN/TiSi_x/n⁺⁺poly‐plug/n⁺ channel layer/Si contact structures, contact resistance—the most important electrical parameter for the diffusion barrier in the bottom electrode structure of capacitors—was found to be as low as 5 kohm, even after annealing up to 750 °C. When the RTN film was inserted as a glue layer between the bottom Pt electrode layer and the TiN barrier film in the chemical vapor deposited (Ba,Sr)TiO₃ (CVD–BST) simple stack‐type structure, the RTN glue layer was observed to be thermally stable to temperatures 150 °C higher than that to which the TiN glue layer is stable. Moreover, the capacitance of the physical vapor deposited (PVD)–BST simple stack‐type structure adopted TiN glue layer initially degraded after annealing at 500 °C, and, thereafter, completely failed. In the case of the RTN and RTO/RTN glue layers, however, the capacitance continuously increased up to 550 °C. Thus, the new RTN and RTO films, which act as diffusion barriers to oxygen, are very promising materials for achieving high‐density capacitors.     

Sensing and gait planning of quadruped walking and climbing robot for traversing in complex environment

In this paper, a general study on improving adaptability of quadruped walking and climbing robot in complex environment is presented. First, a sensing system composed of range and gyroscope sensors in a novel arrangement is developed. By combining the sensing signals and the internal state of the robot, the surface geometry of the environment is sufficiently reconstructed in real-time. Secondly, a planning algorithm for the robot to overcome the reconstructed environment is conducted. Based on the reshaped surface, the planning algorithm not only provides the exact body trajectory and foot positions but also the adaptability of the robot in a specific environment. A method to improve the adaptability of the walking and climbing robot is also introduced. Thanks to the adherent ability of the robot, the center of gravity of the robot is allowed to move outside the support polygon to increase the reach-ability of the next swing leg. Finally, the effectiveness of the proposed approach is verified by the performances of the experiments in complex environments using a quadruped walking and climbing robot named MRWALLSPECT IV.     

A 1.2 V 83 dB DR single‐ended input SC ΔΣ modulator including a large‐swing analog buffer for portable ECG applications

This study proposes a subsystem consisting of an analog buffer and a single‐ended input to a fully differential ΔΣ modulator to obtain low‐power consumption for portable electrocardiogram applications. With the proposed subsystem, the need for an inverting amplifier is avoided, and low‐power consumption is achieved. The ΔΣ modulator with a second order, 1 bit, and cascade of integrators feedforward structure consumes a low power, in which an inverting and a non‐inverting path implement a single‐ended input to fully‐differential signals. A double sampling technique is proposed for a digital‐to‐analog converter feedback circuit to reduce the effect of the reference voltage, reduce the amplifier requirements, and obtain low‐power consumption. Input‐bias and output‐bias transistors working in the weak‐inversion region are implemented to obtain an extremely large swing for the analog buffer. At a supply voltage of 1.2 V, signal bandwidth of 250 Hz, and sampling frequency of 128 kHz, the measurement results show that the modulator with a buffer achieves a 77 dB peak signal‐to‐noise‐distortion ratio, an effective‐number‐of‐bits of 12.5 bits, an 83 dB dynamic range, and a figure‐of‐merit of 156 dB. The total chip size is approximately 0.28 mm² with a standard 0.13 µm Complementary Metal‐Oxide‐Silicon (CMOS) process. Copyright © 2016 John Wiley & Sons, Ltd.