Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
LUTNet-RC: Look-Up Tables Networks for Reservoir Computing on an FPGA
Get PDFWe propose look-up tables networks-based reservoir computing (LUTNet-RC). This work is the first trial of applying LUTNets to RC. LUTNet-RC consists of a LUT-based reservoir layer and a non-LUT-based output layer. LUTNets have disadvantages such as limited sparse connectivity and weights cannot be changed after implementation. However, when applied to a reservoir layer of RC (LUT-based reservoir layer), these disadvantages are eliminated, because this layer works with sparse connectivity and the weights are fixed, so only the advantage of small circuit resources is obtained. For the LUT-based reservoir layer, we propose and model a multi-bit weight reservoir, modifying the conventional binarized reservoir to improve calculation accuracy. In the case of LUTNets, the proposed multi-bit weight reservoir can be implemented without the increase in utilized circuit resources because LUTNets focus only on the input-output relationship on neurons. Additionally, we propose a speed-up method in the output layer with time division calculation, which compares the current network state with previous states and then calculates only status-changed neurons. As a result, we implement a LUTNet-RC with 1500 reservoir neurons on a field-programmable gate array (KR260) running at 100MHz. The utilized circuit resources are dominated by LUTs, which use approximately 26% of the total amount of LUTs. The LUTNet-RC can infer more than 10 6 data per second. We also verify the LUTNet-RC performance using nonlinear auto-regressive moving average 10 (NARMA10) and the performance is comparable to conventional works. We conclude that the LUTNet-RC is one of the highest-performance RC on an FPGA.2023 International Conference on Field Programmable Technology (ICFPT), 12-14 December 2023, Yokohama, Japanjournal articl
Fused Deposition Modelingの積層条件がスナップフィットの機械特性と締結音に与える影響
Get PDF九州工業大学九州工業大学博士学位論文（要旨）学位記番号:情工博甲第392号 学位授与年月日:令和6年3月25日thesi
アミノ酸配列情報のみからの深層学習によるヒト-ウイルスタンパク質間相互作用の予測に関する研究
Get PDF九州工業大学九州工業大学博士学位論文（要旨）学位記番号:情工博甲第396号 学位授与年月日:令和6年3月25日thesi
Controlling adsorption of two dyes on TiO2 surface to improve the efficiency of see-through dye-sensitized solar cells
No full textTransparent dye-sensitized solar cells (DSSCs) were fabricated utilizing judiciously selected blue-absorbing (D-131) and red-absorbing (SQ-140) two sensitizing dyes. These dyes exhibit complementary light absorption along with the least optical absorption in the 500 nm-600 nm wavelength region. This is helpful to impart not only high average visible transparency (AVT) but also high photoconversion efficiency (PCE). D-131 exhibits not only an approximately 1.5 times higher adsorption speed but also an approximately 4 times stronger binding on the TiO2. Stepwise dye adsorption on TiO2 was conducted to suppress unwanted and non-favourable inter-dye interactions. It has been demonstrated that when the SQ-140 adsorbed substrate was subjected to further dye adsorption by the D-131, there was a replacement of the SQ-140 by the D-131 reaching the optimal presence of both of the dyes in 20–30 min giving the best PCE of 4.5 %. However, in the opposite sequence of dye adsorption, there was no replacement of D-131 by SQ-140, rather SQ-140 was adsorbed in the available surface on the mesoporous TiO2 adsorbed with D-131. In this 2nd case, the optimal presence of both dyes is attained in 40 min of adsorption by SQ-140 giving the best PCE of 5.45 %. Transparent DSSCs fabricated using stepwise co-sensitized photoanodes demonstrated an AVT of 47.5 % and 26.5 % with thinner (single-layer) and thicker (double-layer) mesoporous TiO2, respectively.journal articl
Development of an information retrieval and exploration system for identifying user's domain of interest
Get PDF5th International Symposium on Neuromorphic AI Hardware, March 1-2, 2024, Kyushu Institute of Technologyconference pape
マイクロ波照射下における化学反応に関する研究
Get PDF九州工業大学博士（情報工学）第1章 緒論| 第2章 マイクロ波照射下における化学反応の エンタルピー的マイクロ波効果係数とエントロピー的マイクロ波効果係数による速度論的な評価| 第3章 マイクロ波照射下における化学反応のための新しい装置| 第4章 誘電体試料の複素誘電率の簡易的測定| 第5章 生細胞に対する生化学的なマイクロ波効果| 第6章 総括九州工業大学博士学位論文 学位記番号: 情工博甲第391号 学位授与年月日:令和6年3月25日令和5年度doctoral thesi
水中音響通信におけるパワー変調により通信量を2倍にする二次元OFDMシステム
Get PDF九州工業大学博士（工学）1 Introduction and Related Work| 2 Underwater Acoustic Channel and OFDM-SPM| 3 Results and Discussions| 4 Conclusion and Future WorkUnderwater communication is one of the most important and difficult challenges facing researchers due to the high attenuation of the signal, communication with the surface because of the harsh medium of water, and data transmission performance degradation because of various effects. Underwater acoustic communication (UWA) has a low data rate, which describes the disadvantage of this type of communication. In addition, it has a low bandwidth range and high latency but has a long transmission range as an advantage.
Multicarrier wireless transmission systems increase the data rate by sending the data using more than one carrier. We proposed a noncoherent orthogonal frequency division multiplexing (OFDM) method to increase the data rate in UWA communication systems.
In addition, doubling the data rate in the OFDM using Subcarrier Power Modulation (OFDM-SPM) system can save half of the bandwidth. The MATLAB simulation program was used to implement the system in the underwater acoustic environment to increase its throughput. The proposed design uses Differential Phase Shift Keying (DPSK) with power control, and the data stream is transmitted through two-dimensional modulation schemes, the DPSK, and the power level of each subcarrier in the OFDM system with cyclic prefix (CP). The underwater channel was designed using a Rician fading multipath with a spreading loss formula as a function of distance and frequency.
The research is based on three steps, the first of which is research on the underwater communication environment, simulating this environment with the proposed method to double the data rate, and implementing this method through practical experiments.九州⼯業⼤学博⼠学位論⽂ 学位記番号：生工博甲第485号 学位授与年⽉⽇: 令和6年3⽉25⽇令和5年度doctoral thesi
通信トラヒックを用いたIoTデバイスの機能実行制御システムに関する研究
Get PDF九州工業大学九州工業大学博士学位論文（要旨）学位記番号: 生工博甲第496号 学位授与年月日:令和6年9月25日thesi
Distinct locomotor adaptation between conventional walking and walking with a walker
Get PDFRolling walkers are common walking aids for individuals with poor physical fitness or balance impairments. There is no doubt that rolling walkers are useful in assisting locomotion. On the other hand, it is arguable that walking with rolling walkers (WW) is effective for maintaining or restoring the nervous systems that are recruited during conventional walking (CW). This is because the differences and similarities of the neural control of these locomotion forms remain unknown. The purpose of the present study was to compare the neural control of WW and CW from the perspective of a split-belt adaptation paradigm and reveal how the adaptations that take place in WW and CW would affect each other. The anterior component of the ground reaction (braking) forces was measured during and after walking on a split-belt treadmill by 10 healthy subjects, and differences in the peak braking forces between the left and right sides were calculated as the index of the split-belt adaptation (the degree of asymmetry). The results demonstrated that (1) WW enabled subjects to respond to the split-belt condition immediately after its start as compared to CW; (2) the asymmetry movement pattern acquired by the split-belt adaptation in one gait mode (i.e., CW or WW) was less transferable to the other gait mode; (3) the asymmetry movement pattern acquired by the split-belt adaptation in CW was not completely washed out by subsequent execution in WW and vice versa. The results suggest unique control of WW and the specificity of neural control between WW and CW; use of the walkers is not necessarily appropriate as training for CW from the perspective of neural control.journal articl