This study presents an autoencoder-embedded optimization(AEO)algorithm which involves a bi-population cooperative strategy for medium-scale expensive problems(MEPs).A huge search space can be compressed to an informat...
详细信息
This study presents an autoencoder-embedded optimization(AEO)algorithm which involves a bi-population cooperative strategy for medium-scale expensive problems(MEPs).A huge search space can be compressed to an informative lowdimensional space by using an autoencoder as a dimension reduction *** search operation conducted in this low space facilitates the population with fast convergence towards the *** strike the balance between exploration and exploitation during optimization,two phases of a tailored teaching-learning-based optimization(TTLBO)are adopted to coevolve solutions in a distributed fashion,wherein one is assisted by an autoencoder and the other undergoes a regular evolutionary ***,a dynamic size adjustment scheme according to problem dimension and evolutionary progress is proposed to promote information exchange between these two phases and accelerate evolutionary convergence *** proposed algorithm is validated by testing benchmark functions with dimensions varying from 50 to *** indicated in our experiments,TTLBO is suitable for dealing with medium-scale problems and thus incorporated into the AEO framework as a base *** with the state-of-the-art algorithms for MEPs,AEO shows extraordinarily high efficiency for these challenging problems,t hus opening new directions for various evolutionary algorithms under AEO to tackle MEPs and greatly advancing the field of medium-scale computationally expensive optimization.
Memory modeling, aimed at predicting the memory states of learners throughout their learning process, has become an integral component in online learning systems. This is particularly crucial for applications such as ...
详细信息
ISBN:
(纸本)9783031643118;9783031643125
Memory modeling, aimed at predicting the memory states of learners throughout their learning process, has become an integral component in online learning systems. This is particularly crucial for applications such as spaced repetition scheduling and knowledge tracking. However, existing machine learning-based memory modeling methods encounter challenges with weak supervision signals and sparse interaction data. To tackle these issues, we propose a novel approach named LSTM autoencoder Collaborative Filtering (LACF) for modeling learner memory. Our model utilizes an LSTM autoencoder to extract temporal features from user-item interaction sequences. Additionally, a collaborative filtering module, based on the deepFM architecture, is employed to effectively learn interactions between different features, facilitating comprehensive low-order and high-order feature combinations. Experimental results demonstrate that LACF significantly outperforms existing state-of-the-art memory modeling methods, offering enhanced accuracy and precision in predictions.
The volume of genomic data being generated is growing due to the technological breakthroughs in genome sequencing and the economic affordability of sequencing. Consequently, the need for compression techniques that ar...
详细信息
ISBN:
(纸本)9798350385939;9798350385922
The volume of genomic data being generated is growing due to the technological breakthroughs in genome sequencing and the economic affordability of sequencing. Consequently, the need for compression techniques that are specifically designed for genomic data is rising. Deep learning provides robust methods for reducing the data size and unlocking the full potential of genomic information. The proposed method minimizes storage requirements by applying a 2-D convolutional neural network autoencoder, which learns spatial and sequential redundancies in the sequence to compress the data losslessly. In contrast to conventional compression methods, which handle data as a one-dimensional sequence, the proposed approach uses the spatial structure of the data to produce more compact representations. We evaluated the proposed method with various compression approaches, including state-of-the-art DNA sequence compression algorithms, on two homo sapiens genomes. Experimental results indicate that the proposed method can compress the genomic data effectively, with an improvement of 31.3% compression over the best existing method.
This paper proposes a novel unsupervised method for muscle fatigue recognition. It uses a convolutional autoencoder to extract time-frequency EMG features and clusters the features into fatigue and non-fatigue groups....
详细信息
ISBN:
(纸本)9783031625015;9783031625022
This paper proposes a novel unsupervised method for muscle fatigue recognition. It uses a convolutional autoencoder to extract time-frequency EMG features and clusters the features into fatigue and non-fatigue groups. Experimental results show that the proposed method is more effective in discriminating muscle fatigue compared to conventional approaches. In addition, the clusters provide an effective way to determine a threshold for identifying muscle fatigue.
Survival prediction for esophageal squamous cell cancer (ESCC) is crucial for doctors to assess a patient's condition and tailor treatment plans. The application and development of multi-modal deep learning in thi...
详细信息
ISBN:
(纸本)9798350313345;9798350313338
Survival prediction for esophageal squamous cell cancer (ESCC) is crucial for doctors to assess a patient's condition and tailor treatment plans. The application and development of multi-modal deep learning in this field have attracted attention in recent years. However, the prognostically relevant features between cross-modalities have not been further explored in previous studies, which could hinder the performance of the model. Furthermore, the inherent semantic gap between different modal feature representations is also ignored. In this work, we propose a novel autoencoder-based deep learning model to predict the overall survival of the ESCC. Two novel modules were designed for multi-modal prognosis-related feature reinforcement and modeling ability enhancement. In addition, a novel joint loss was proposed to make the multi-modal feature representations more aligned. Comparison and ablation experiments demonstrated that our model can achieve satisfactory results in terms of discriminative ability, risk stratification, and the effectiveness of the proposed modules.
Side-channel data compression techniques are designed to reduce the dimensionality of input data while preserving critical information, thereby improving the efficiency of side-channel analysis. Traditional methods of...
详细信息
ISBN:
(纸本)9798350377040;9798350377033
Side-channel data compression techniques are designed to reduce the dimensionality of input data while preserving critical information, thereby improving the efficiency of side-channel analysis. Traditional methods of side-channel analysis can be broadly classified into two categories: peak extraction techniques and data integration techniques, based on the scale and principles of the input data. In this study, we present a novel side-channel data compression technique utilizing convolutional autoencoder networks. We apply this method to power-side channel data collected from an SM4 encryption circuit implemented on a secure chip. Experimental results show that our proposed technique surpasses conventional methods in Test Vector Leakage Assessment (TVLA) analysis and achieves a 35% improvement in efficiency in Measurements to Disclosure (MTD) analysis, while maintaining the same compression rate. These results underscore the effectiveness and potential of our approach in enhancing side-channel analysis.
Denoising autoencoders for signal processing applications have been shown to experience significant difficulty in learning to reconstruct radio frequency communication signals, particularly in the large sample regime....
详细信息
ISBN:
(纸本)9798350374247;9798350374230
Denoising autoencoders for signal processing applications have been shown to experience significant difficulty in learning to reconstruct radio frequency communication signals, particularly in the large sample regime. In communication systems, this challenge is primarily due to the need to reconstruct the modulated data stream which is generally highly stochastic in nature. In this work, we take advantage of this limitation by using the denoising autoencoder to instead remove interfering radio frequency communication signals while reconstructing highly structured FMCW radar signals. More specifically, in this work we show that a CNN-layer only autoencoder architecture can be utilized to improve the accuracy of a radar altimeter's ranging estimate even in severe interference environments consisting of a multitude of interference signals. This is demonstrated through comprehensive performance analysis of an end-to-end FMCW radar altimeter simulation with and without the convolutional layer-only autoencoder. The proposed approach significantly improves interference mitigation in the presence of both narrow-band tone interference as well as wideband QPSK interference in terms of range RMS error, number of false altitude reports, and the peak-to-sidelobe ratio of the resulting range profile. FMCW radar signals of up to 40,000 IQ samples can be reliably reconstructed.
Drone surveillance radars are dependent on reliable classification of targets for useful operation. Machine learning-based approaches, such as those based on deep learning, have been helpful for advancing such sensors...
详细信息
ISBN:
(纸本)9798350329216;9798350329209
Drone surveillance radars are dependent on reliable classification of targets for useful operation. Machine learning-based approaches, such as those based on deep learning, have been helpful for advancing such sensors' operation in real unpredictable environments. However, with labelled data being scarce resource in radar domain, machine learning-based approaches inherently carry several performance uncertainties. This is true even with approaches that involve domain translation or transfer learning, for example, with models evolved from optically trained models. In this paper, a comparison of the classification effectiveness of two different types of machine learning based approaches is performaned, namely, a convolutional neural network (CNN) pretrained with optical data, and an autoencoder-based model trained with real-world radar-only data. This comparison unlike others in the literature compares supervised and unsupervised pretraining techniques. Our results show that the unsupervised approach can outperform the resourcedemanding supervised approach based on transfer learning. Furthermore, autoencoder pretraining repeated with synthetic micro-Doppler data yielded near identical classification results, which paves the possibility to utilize greater amounts of synthetic data for pretraining deep learning models. A brief inspection of the latent distribution of the simple symmetric, unregularized autoencoder confirms minor preservation of features in the learned representation.
Feature extraction is a critical task in the design of intelligent condition monitoring (CM) systems. In most cases, expert knowledge is required. This study investigates feature extraction for bearing condition monit...
详细信息
ISBN:
(纸本)9798350387605;9798350387599
Feature extraction is a critical task in the design of intelligent condition monitoring (CM) systems. In most cases, expert knowledge is required. This study investigates feature extraction for bearing condition monitoring using artificial intelligence (At). A hierarchical extreme learning machine (HELM) is used to design a condition monitoring system (CMS) for bearings that can be used for fault classification and anomaly detection. Acceleration signals are processed using an autoencoder to extract features and anomalies were detected or faults were classified depending on the task of the CMS. Almost-parameter-free machine learning-based classifiers have been used to solve the classification task. In general, the study employed a (minimum) distance classifier (MiDC), and one -class k-nearest neighbors (OC-k-NN). This was done to show that deep learning -based feature extraction provides excellent spatial separation and facilitates classification with minimal training effort. The study compared the distance based classifiers with the performance of several artificial intelligence-based approaches, such as support vector machine (SVM) one-class support vector machine (DC-SVM), isolation forests (IF), and multi -layer perceptron (MLP). The proposed method was evaluated using both the CWRU dataset and an own dataset. The proposed method achieved Fl scores above 0.98 for both anomaly detection and fault classification with the CWRU dataset and an own daset.
An evaporator is a critical component in waste heat recovery systems, particularly within organic Rankine cycle technologies. In this paper, an approach for the simulation of two-phase evaporators is presented. This m...
详细信息
ISBN:
(纸本)9798350364965;9798350364972
An evaporator is a critical component in waste heat recovery systems, particularly within organic Rankine cycle technologies. In this paper, an approach for the simulation of two-phase evaporators is presented. This method aims at dimension reduction and prediction of temporal measurements, leveraging an autoencoder integrated with long short-term memory neural networks. The autoencoder-LSTM integration effectively reduces dimensions by compressing data representations and enhances temporal predictions, capturing dynamic system behaviors efficiently. The methodology effectiveness in predicting system dynamics is demonstrated, including a case study on high-fidelity data generation. In the absence of explicit system equations and with only snapshot matrices available, this method is preferred over projection-based model order reduction methods for its enhanced capability to capture and predict complex, non-linear temporal dynamics directly from data. The potential of integrating deep learning techniques with conventional engineering simulations to address the challenges of complex energy systems is highlighted. The results indicate that architectures with varying numbers of neurons in the examined bottleneck layer can accurately reconstruct and predict the original system's transient and steady-state behaviors with an error rate below 8%.
暂无评论