This paper focuses on speech recognition algorithm based on bidirectional long short-term memory network combined with Connected Temporal classification (BiLSTM-CTC), aiming to improve the performance of spoken Englis...
详细信息
ISBN:
(数字)9798331527662
ISBN:
(纸本)9798331527679
This paper focuses on speech recognition algorithm based on bidirectional long short-term memory network combined with Connected Temporal classification (BiLSTM-CTC), aiming to improve the performance of spoken English evaluation system. The algorithm can improve the accuracy and robustness of speech recognition effectively by analyzing the English speech signals after pre-emphasis processing. Firstly, the original speech signal is preweighted to enhance the high frequency part and reduce the frequency resonance effect. Then Mayer frequency cepstrum coefficient (Fbank) is used as feature extraction method to capture the spectral characteristics of speech signals. The experimental results show that the amplitude of energy change of English speech signal after pre-accentuated is stable between-35dB and OdB, and the spectrum distribution is more balanced. This model has a significant improvement in recognition speed and accuracy. Especially in the complex background noise environment, the anti-interference ability of the new model is particularly outstanding, and the false recognition rate is effectively reduced. The speech recognition algorithm based on BiLSTM-CTC proposed in this paper not only optimizes the speech signalprocessing flow, but also greatly improves the overall performance of the recognition system.
Cyber-attacks have surged in intensity and complexity, often outpacing existing detection systems. advanced attack prediction algorithms can help mitigate these threats, reducing their impact and strengthening network...
详细信息
ISBN:
(数字)9798331532543
ISBN:
(纸本)9798331532550
Cyber-attacks have surged in intensity and complexity, often outpacing existing detection systems. advanced attack prediction algorithms can help mitigate these threats, reducing their impact and strengthening network security. While prior research has advanced attack predictability and source estimation, deep graph modeling has yet to integrate such advancements fully. This article introduces a novel method for binary link prediction in cyber-attack networks, leveraging dynamic graph deep learning and temporal graph neural networks (TGN). The approach begins by constructing a time-evolving attack graph based on historical alert data within a specific time window. This dynamic graph is then fed into a TGN, which automatically extracts structural and temporal features. The model predicts the likelihood of future attack links between entities, enabling more effective identification of potential cyber threats before they occur. In this article, three different deep networks were evaluated for predicting attacks. The dynamic graph method demonstrated 10–18% better performance compared to the other methods, improving accuracy in link prediction.
We target the source localization problem using only changing rate of phase difference (CRPD) measurements collected by a moving platform with a long baseline interferometer. Source localization using CRPD measurement...
详细信息
ISBN:
(数字)9789464593617
ISBN:
(纸本)9798331519773
We target the source localization problem using only changing rate of phase difference (CRPD) measurements collected by a moving platform with a long baseline interferometer. Source localization using CRPD measurements is a non-convex problem requiring advanced techniques to be resolved. In our paper, we first form a constrained-weighted least squares (CWLS) problem from the maximum-likelihood (ML) function. Then, we transform the CWLS problem into a semi-definite programming. By dropping the rank-one constraint, we achieve semi-definite relaxation, and the problem becomes convex, which interior-point algorithms can optimally solve. In the simulations, we compare the proposed method to the pseudo-linear approaches, ML solver, and Cramer-Rao lower bound (CRLB). We observe that the proposed method attains the CRLB at low noise levels and outperforms the pseudo-linear approaches while performing similarly to the ML solver.
With the rapid expansion of video services, caching has emerged as a promising solution to reduce retrieval delay and alleviate backhaul pressure. Specifically, caching video content on user devices and satellites is ...
详细信息
ISBN:
(数字)9798350390643
ISBN:
(纸本)9798350390650
With the rapid expansion of video services, caching has emerged as a promising solution to reduce retrieval delay and alleviate backhaul pressure. Specifically, caching video content on user devices and satellites is particularly effective for users without ground base station services. However, addressing the challenge of meeting users' varying video quality demands while reducing content transmission delay is critical. Therefore, this paper investigates a layered caching scheme based on Scalable Video Coding, which encodes videos into different layers to provide various quality levels and caches video content on different nodes. Furthermore, an optimization problem for the probabilistic cache placement scheme in a multi-layer caching model is proposed and solved using the Quasi-Oppositional Whale Optimization Algorithm. The simulation results indicate that, compared to other algorithms and benchmark schemes, the proposed scheme effectively reduces transmission delay.
Quantum computing is at the leading edge of advanced technology that uses quantum phenomena such as entanglement and super position to solve complex problems. One of the main application is to built comparator circuit...
详细信息
ISBN:
(数字)9798331540685
ISBN:
(纸本)9798331540692
Quantum computing is at the leading edge of advanced technology that uses quantum phenomena such as entanglement and super position to solve complex problems. One of the main application is to built comparator circuit to do arithmetic operations at stipulated time. Based on this our research is to propose the design and implementation of one-bit and two-bit quantum comparators using IBM Qiskit, which is formly compared with classical comparator. Also this work represents a comprehensive overview of the techniques required to optimize and design quantum comparator circuits using real quantum simulators. The results obtained from simulation promises the comparison accuracy with classical results. From the circuit designed it is inferred that the hardware complexity is very much reduced compared to classical comparator, which are tabulated in terms of number of quantum gates. The proposed method can further be extended to generate higher order comparator circuits. These results may form the basis for further research and development that will support the advancement of quantum computing and algorithms.
RF (radio frequency) signals are commonly used in modern communication systems, e.g., radio broadcasting, mobile electronic devices, smart meters, satellite transmission, etc., but often face significant challenges su...
详细信息
ISBN:
(数字)9798350367621
ISBN:
(纸本)9798350367638
RF (radio frequency) signals are commonly used in modern communication systems, e.g., radio broadcasting, mobile electronic devices, smart meters, satellite transmission, etc., but often face significant challenges such as signal interference and/or background noise. To gain a deeper understanding of signal extraction methods, in this paper, we perform an initial survey on the existing approaches and general trends in extracting signals of interest from the mixed signals received in cluttered environments, including advanced sensing, filtering, data augmentation, machine learning, and emerging quantum-enhanced signalprocessingalgorithms. We then study a specific machine learning model that employs a Bayesian neural network (BNN) as the encoder, combined with a decoder based on a deep neural network (DNN), a long short-term memory (LSTM) network, or another BNN. Case studies are performed to compare different autoencoder neural network models, and initial testing results show that the BNN's probabilistic parameters can handle input signal uncertainty well and may significantly enhance feature extraction and signal clarity. Planned future research will explore quantum computing methods to enhance RF signal extraction in cluttered environments, and some potential quantum algorithms are also identified in the paper.
Early detection of the diseases is very vital for the well-being and improvement of the crop. Often, it is difficult to detect the disease of the crop upon observation with a naked eye. Professionals in identifying pl...
详细信息
ISBN:
(数字)9798350350654
ISBN:
(纸本)9798350350661
Early detection of the diseases is very vital for the well-being and improvement of the crop. Often, it is difficult to detect the disease of the crop upon observation with a naked eye. Professionals in identifying plant diseases are very less and their availability in each and every region is one of the primary difficulties that the agriculture sector is facing. So, the best solution for this problem is to replace a human with an advanced and well-trained model that can help in accurate detection of the plant disease. Several people have researched in this field. Yet, the results are not always promising and there is lot of scope for improvement. In this project, we made use of Groundnut crop for determining the performance of the disease detection model that we are developing. After referring to various papers, we eventually decided to work on three models- CNN, MobilenetV2 and InceptionResnetV2. Our ultimate aim is to determine the best model among these three for groundnut disease detection.
In the past decades, there has been an extensive research interest in the areas of both waveform diversity/design and advancedsignalprocessingalgorithms departing from the more classical solutions based on Linear F...
详细信息
ISBN:
(纸本)9781728176093
In the past decades, there has been an extensive research interest in the areas of both waveform diversity/design and advancedsignalprocessingalgorithms departing from the more classical solutions based on Linear Frequency Modulated (LFM) pulses and Matched Filters (MF). In the waveform diversity community, especially within the context of spectrum sharing, MIMO and cognitive radars, several waveform optimization and design methodologies have been studied, see [1], [2] and references therein. In parallel to waveform design, several signalprocessing techniques have also been proposed which exploit some kind of prior knowledge and/or iterative algorithms to improve the performance of the more classical MF, such as the Adaptive Pulse Compression(APC) [3], MUSIC [4], CLEAN [5] and Sparse signalprocessing (SSP) [6], [7]. In this paper we present some results and examples to show how the combination of waveform design with SSP can lead to improved performance in radar compared to the more classical approach.
Wireless Sensor Networks (WSNs) are essential for many applications, such as industrial automation, healthcare, and environmental monitoring. As these networks continue to grow, ensuring the security and energy effici...
详细信息
ISBN:
(数字)9798350364590
ISBN:
(纸本)9798350375381
Wireless Sensor Networks (WSNs) are essential for many applications, such as industrial automation, healthcare, and environmental monitoring. As these networks continue to grow, ensuring the security and energy efficiency of data transmission becomes paramount. This research proposes a novel Secure Energy Consumption Technique (SECT) designed to address the simultaneous challenges of securing communication and optimizing energy consumption in WSNs. The SECT integrates advanced encryption algorithms and efficient key management protocols to fortify the security of data transmission within the network. By employing lightweight cryptographic techniques, the proposed solution aims to assault a balance between security and energy efficiency, ensuring that the overhead associated with encryption does not compromise the sensor nodes' restricted energy resources. Key features of the SECT include a dynamic key management system that adapts to the network's changing conditions, minimizing energy consumption while maintaining a robust security posture. Additionally, the technique incorporates energy-aware routing algorithms to optimize the path selection for data transmission, further reducing energy expenditure.
With the rapid development of satellite communication technology, the emergence of interference signals is becoming more and more frequent, which seriously affects the reliability and safety of communication. For this...
详细信息
ISBN:
(数字)9798331541668
ISBN:
(纸本)9798331541675
With the rapid development of satellite communication technology, the emergence of interference signals is becoming more and more frequent, which seriously affects the reliability and safety of communication. For this reason, this paper designs a satellite communication interference signal recognition system based on artificial intelligence, which adopts a modularized design and covers multiple aspects such as signal acquisition, pre processing, feature extraction and classification recognition. The system realizes high-fidelity signal acquisition through a high dynamic range broadband receiver, uses FPGA to carry out real-time data preprocessing, and combines deep learning technology to adaptively extract interference signal features. The classification and recognition module incorporates a variety of advancedalgorithms to significantly improve the accuracy of detection and recognition. Experimental results show that the system designed in this paper exhibits excellent detection performance under different signal-to-noise ratio conditions, and the recognition accuracy reaches $\mathbf{9 7. 5 \%}$ on average, which is 5.7 percentage points higher than the traditional SVM-based algorithm. This research provides an effective guarantee for the anti-jamming capability of satellite communication system, which has important application value and promotion potential.
暂无评论