In this work, the classification of walking direction based on ultrasonic signals has been examined for entrance counting. Feed-forward and recurrent neural network architectures as well as simpler machine learning te...
详细信息
ISBN:
(数字)9781665490627
ISBN:
(纸本)9781665490627
In this work, the classification of walking direction based on ultrasonic signals has been examined for entrance counting. Feed-forward and recurrent neural network architectures as well as simpler machine learning techniques have been investigated and compared with classical signal processing techniques. Using only a single ultrasonic receiver, the focus was set on the development of a hardware-efficient system concept. Different ultrasonic measurement methods in time and frequency domain have been compared withthe perspective of a holistic energy optimization. the analysis of the system's hardware efficiency was completed by an estimation of algorithmic latency, energy and storage consumption based on the arithmetic of the classification algorithms. All algorithms showed an estimated energy consumption of less than 10 mu J for a single inference on a state-of-the-art implementation of an ARM (R) Cortex (R) M4F micro-controller, which was found to be negligible compared to the energy of the measurement principle. Compared to other sensor types and multi-sensor systems, a state-of-the-art test accuracy of 99.72% could be achieved for differentiating between the two entrance directions of a present person and the absence of a person.
LIBROSA is a powerful Python audio data processing library introduced in recent years. Based on LIBROSA provided source codes, two types of feature data extraction algorithms are analyzed in detail, including MEL powe...
详细信息
ISBN:
(数字)9798350352719
ISBN:
(纸本)9798350352726
LIBROSA is a powerful Python audio data processing library introduced in recent years. Based on LIBROSA provided source codes, two types of feature data extraction algorithms are analyzed in detail, including MEL power spectrum and CHROMA syllable spectrums. By each algorithm, 128, 12 floating point numbers are extracted as the feature data of their corresponding spectrums. In addition, from multi perspectives of basic music theory, mathematical analysis, signal processing, computer programming and so on, deeply discusses LIBROSA library functions involved Discrete Fourier Transform (DFT), Triangular filter, Gaussian filter, and other programming technologies. the DFT algorithm, time domain window function, syllable feature data extraction, band-pass filter code analysis and other contents discussed have important reference value in data processing, deep learning and other fields of underlying model creation, source codes development and so on
An interesting modern technology called cognitive radio creates new opportunities for the effective utilization of the spectrum. Deep Learning (DL) techniques rely on experimentally recorded data and, when trained pro...
详细信息
ISBN:
(纸本)9781665456456
An interesting modern technology called cognitive radio creates new opportunities for the effective utilization of the spectrum. Deep Learning (DL) techniques rely on experimentally recorded data and, when trained properly with a wide range of data, may effectively recognize the radio settings, adapt to different environments, and constantly provide a great performance. Using a variety of signal processing (SP) features, we compare the performance of various deep neural network (DNN) models for spectrum sensing (SS) in this paper. the features that are taken into consideration are differential entropy, energy, Lp-norm and geometric power. Conventional DNN are trained to perform spectrum sensing (SS) in congnitive radio (CR) with two different models of noise. In one noise model we take experimentally recorded data from an unoccupied frequency modulation broadcast channel and in another noise model we consider generalized Gaussian noise (GGN). through thorough tests based on real-world collected datasets, we find that ResNet and Multilayer perceptron (MLP) architectures provide the most effective result in perspective of likelihood of detection of primary user, for a specific preset value of false-alarm probability.
processing-in-memory (PIM) substrates can perform parallel computation directly within the memory array. As a result, throughput performance and energy efficiency can reach unprecedented levels, however, there are lim...
详细信息
Backdoor attacks pose significant threats to Natural Language processing (NLP) models. Various backdoor defense methods for NLP models primarily function by identifying and subsequently manipulating backdoor triggers ...
详细信息
Nowadays, it is common in many disciplines and application fields to collect large volumes of data characterized by a high number of features. Such datasets are at the basis of modern applications of supervised Machin...
详细信息
Vortex flowmeters occupy an important position in the field of flow measurement. their advantages include having no moving mechanical parts, the ability to adapt to various media, and low-pressure loss. After the vort...
详细信息
ISBN:
(数字)9798331542887
ISBN:
(纸本)9798331542894
Vortex flowmeters occupy an important position in the field of flow measurement. their advantages include having no moving mechanical parts, the ability to adapt to various media, and low-pressure loss. After the vortex signal detected by the piezoelectric element is amplified and filtered through a series of analog and digital circuits, an accurate flow value can be obtained. To enhance the dynamic response performance of the vortex flowmeter, a multi-channel parallel signal processing method with 1/f 2 amplitude-frequency characteristic modulation has been designed. this method employs filtering units with a 1/f 2 attenuation characteristic (-40 dB/dec) combined withthe physical relationship of the vortex signal's amplitude-frequency characteristics to expand the measurement range. the measurement channels are divided into several parallel channels, and a fast channel selection method is designed to effectively address the slow dynamic response speed of traditional vortex flowmeter signal processing methods. the results show that the vortex flowmeter integrated withthe four-channel algorithm achieves 0.5-grade accuracy and exhibits superior dynamic response performance compared to Yokogawa flowmeters.
In an era inundated with high-dimensional consumer data, the need for advanced hyper-dimensional pattern analysis poses a significant computational challenge. this research pioneers the use of quantum computing to rev...
详细信息
Photonic integrated circuits have potential to advance artificial neural networks by providing ultra-fast computing capabilities with high power efficiency. In present conventional optical neural network architectures...
详细信息
this research designed and implemented a leukemia diagnostic system targeted for a real clinical environment based on deep learning approaches. the dataset consists of 15,135 total images. this research develops four ...
详细信息
ISBN:
(纸本)9781450399067
this research designed and implemented a leukemia diagnostic system targeted for a real clinical environment based on deep learning approaches. the dataset consists of 15,135 total images. this research develops four independent models (VGG19 1 epoch, VGG19 30 epochs, ResNet50 1 epoch, and ResNet50 30 epochs), two Hybrid models (trained at 1 epoch and 30 epochs), and four Ensemble models (two Ensemble models of VGG19 and ResNet50 and two Ensemble models with an additional Hybrid model). All models are pre-trained on ImageNet. By using transfer learning, the models were fine-tuned (further trained) on the leukemia domain at a much greater speed as the existing layers will have benefitted from the pre-training done on ImageNet. this research indicates that Hybrid models can help improve predictive capabilities by leveraging different feature patterns extracted from running images through two different architectures. Meanwhile, Ensemble models will take the prediction votes from multiple final model outputs to further incorporate different model capabilities and also help generalize. Among all independent models, the best model is ResNet50 30 epochs, which achieved an accuracy of 84%. Among the Hybrid models, the best model is Hybrid 30 epochs, which achieved an accuracy of 84%. Ensemble 4 (Hybrid 30 epochs, VGG19 1 epoch, and ResNet50 1 epoch) achieved an accuracy of 86%, which is 2% better than the second-best model, Ensemble 2. In principle, the diagnostic system developed in this research can be applied in other medical diagnostic applications.
暂无评论