Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
Non-gain asymmetry for dissipativity framework for positive systems exhibiting interior equilibria
Get PDFThis paper aims to formulate a new approach to dissipative modeling of nonlinear positive systems with interior equilibria. Processing positivity separately from stability becomes unnecessary. For asymmetric state spaces on which systems evolve, this paper proposes a supply rate of non-gain-type asymmetry and develops formulas of their aggregation, which is not as trivial as aggregating L 2 gain and passivity supply rates. The novelty of their proofs is that the involved treatment of asymmetric functions is reduced to the evaluation of directed graphs carrying balance information between variables. The dissipativity on asymmetric spaces and the aggregation formulas qualify a single logarithmic storage function to establish stability and positivity simultaneously for an interconnected system. The usefulness is illustrated by examples.journal articl
LGNMNet-RF: Micro-Expression Detection Using Motion History Images
Get PDFMicro-expressions are very brief, involuntary facial expressions that reveal hidden emotions, lasting less than a second, while macro-expressions are more prolonged facial expressions that align with a person’s conscious emotions, typically lasting several seconds. Micro-expressions are difficult to detect in lengthy videos because they have tiny amplitudes, short durations, and frequently coexist alongside macro-expressions. Nevertheless, micro- and macro-expression analysis has sparked interest in researchers. Existing methods use optical flow features to capture the temporal differences. However, these optical flow features are limited to two successive images only. To address this limitation, this paper proposes LGNMNet-RF, which integrates a Lite General Network with MagFace CNN and a Random Forest classifier to predict micro-expression intervals. Our approach leverages Motion History Images (MHI) to capture temporal patterns across multiple frames, offering a more comprehensive representation of facial dynamics than optical flow-based methods, which are restricted to two successive frames. The novelty of our approach lies in the combination of MHI with MagFace CNN, which improves the discriminative power of facial micro-expression detection, and the use of a Random Forest classifier to enhance interval prediction accuracy. The evaluation results show that this method outperforms baseline techniques, achieving micro-expression F1-scores of 0.3019 on CAS(ME)2 and 0.3604 on SAMM-LV. The results of our experiment indicate that MHI offers a viable alternative to optical flow-based methods for micro-expression detection.journal articl
A Study on Action Selection Probability Model Considering Selection Bias in Reinforcement Learning-Based Behavior Modeling
Get PDFIn this study, we propose an action selection probability model that takes into account the case where there is a certain bias in the action to be selected in reinforcement learning-based action modeling. In the proposed action selection probability model, the softmax function is shifted in parallel when calculating the action selection probability, assuming that factors other than reward influence the selection of actions. Specifically, parallel shift is achieved by adding a certain bias to the difference of action values in each state for calculating the action selection probability. In the proposed method, this bias value is determined based on maximum likelihood estimation in addition to the learning rate and inverse temperature in conventional reinforcement learning models, respectively. In order to confirm the effectiveness of the proposed method, we artificially generated data that is likely to take a certain action independent of the reward using a two-armed bandit problem, which is a type of benchmarking, and compared the likelihood of each model in the conventional and proposed methods using this data. The results showed that the likelihood of the proposed method was significantly higher than that of the conventional method.journal articl
A Hippocampus-Inspired Memory Model with an Accumulation Function Through Place Representation Planes for Digital VLSI Implementation
Get PDFAcquisition of episodic memory is important to handle environment-specific information for home service robots. Such privacy data should be processed on the edge computer mounted on the robot with low energy consumption. A hippocampus-inspired model and its digital very large-scale integration (VLSI) implementation were proposed to acquire episodic memory [1]. The previously proposed method represents episodes as locations of objects. We propose an expanded model and its VLSI design with place representation planes and a path plane as shown in Figure 1 (a). The path plane represents the location histories of robot movements to memorize object locations. Location histories are attenuated as time elapsed and accumulated to represent a path. In the VLSI implementation, the accumulation operation is performed in parallel for each processing group; i.e., ! × × # planes are accumulated in ! × # steps, where ! denotes the number of planes for the history of locations, " and # denotes the number of cells in -axis and -axis respectively, as shown in Figure 1 (b). We simulated the circuit using Vitis HLS 2022.2 to verify the behavior of the accumulation function."journal articl
Enhancing Campus Mobility: Simulated Multi-Objective Optimization of Electric Vehicle Sharing Systems Within an Intelligent Transportation System Frameworks
Get PDFThis research optimizes an electric vehicle (EV) sharing system for a university campus, focusing on different demand patterns and peak times within an Intelligent Transportation System (ITS) framework. The main objectives are to reduce the number of unserved demands and operational costs. A simulation model was developed in MATLAB, utilizing the Non-dominated Sorting Genetic Algorithm (NSGA-II), a powerful multi-objective optimization technique that balances conflicting objectives to achieve the best trade-offs for operational efficiency. In addition to conventional decision variables, dynamic dual relocation thresholds and charge levels are introduced as decision variables to enhance optimization. The study compares two scenarios: Equally Distributed Demand (EDD) and Non-Equally Distributed Demand (NEDD), customized for the University Putra Malaysia (UPM) campus. Findings indicate that the NEDD scenario, which concentrates on specific demand areas, effectively decreases unserved demands and operational costs. Additionally, a station-specific approach expanded the solution space, improving adaptability and resulting in notable reductions in operational costs and smaller but meaningful improvements in unserved demands, especially during peak periods. By setting station-specific relocation thresholds and charge levels, resources were deployed efficiently, minimizing unnecessary relocations. The use of dynamic values for dual relocation thresholds and charge-to-work levels further optimized the process, reducing operational costs significantly, with a lesser impact on unserved demands across both scenarios. This research offers valuable insights into the implementation of EV sharing systems in educational institutions, emphasizing the advantages of focused resource allocation and the integration of dynamic decision variables.journal articl
Design of Asynchronous Information Literacy Lecture to Promote Students’ Self-directed Learning
Get PDFVarious educational materials have become available on the internet in recent years, allowing students to learn freely according to their interests. Consequently, we believe that it is necessary to respond to students’ learning styles as students increasingly seek these materials and continuously learn on their own. In this study, we designed a model for asynchronous lectures that prioritized student autonomy to promote self-directed learning, and we applied this model to information literacy lectures. We observed autonomous learning activities among some students. However, in this lecture model, once students’ submissions of assignments was delayed, this pattern tended to persist. In this paper, we confirm that by addressing the shortcomings of this model in solving this problem, we can reapply it to information literacy lectures. Based on the analysis of student learning activities in this course, we confirmed the sustainability of autonomous learning activities.journal articl
Phase binarization in mutually synchronized bias field free spin-Hall nano-oscillators for reservoir computing
Get PDFMutually coupled spin-Hall nano-oscillators (SHNO) can exhibit a binarized phase state, offering pathways to realize Ising machines and efficient neuromorphic hardware. Conventionally, phase binarization is achieved in coupled identical SHNOs via injecting an external microwave at twice the oscillator frequency in the presence of a strong biasing magnetic field. However, this technology poses potential challenges of higher energy consumption and complex circuit design. Moreover, differences in the individual characteristic frequencies of SHNOs resulting from fabrication-induced mismatch in SHNO dimensions may hinder their mutual synchronization. Addressing these challenges, we demonstrate purely dc current-driven mutual synchronization and phase binarization of two nonidentical nanoconstriction SHNOs without biasing magnetic field and microwave injection. We thoroughly investigate these phenomena and underlying mechanisms using micromagnetic simulation. We show how the localized fundamental mode of the spin wave emerging from the magnetization auto-oscillation reinforces the mutual synchronization, while the second-harmonic spin wave induces the phase binarization in the coupled SHNO pair. We further demonstrate the bias field free synchronized SHNO pair efficiently performing a reservoir computing benchmark learning task: sin- and square-wave classification, with 100% accuracy, utilizing the current-tunable phase binarization phenomenon. Our results showcase promising magnetization dynamics of coupled bias field free SHNOs for future computing applications.journal articl
