Collaborative Filtering (CF) has been a highly effective approach widely utilized in Recommender Systems, leveraging user-item interactions. Among CF methods, Matrix Factorization (MF) has maintained a prominent statu...
详细信息
ISBN:
(纸本)9783031770425;9783031770432
Collaborative Filtering (CF) has been a highly effective approach widely utilized in Recommender Systems, leveraging user-item interactions. Among CF methods, Matrix Factorization (MF) has maintained a prominent status as a state-of-the-art model, delivering personalized item recommendations based on user preferences. Recent advancements have integrated Deep Learning techniques into MF, particularly to capture nonlinear item representations and enhance recommendation accuracy. However, many of these models often overlook user-item features and fail to offer understandable explanations for their recommendations. This absence of explanation mechanisms undermines user trust in recommendations, reducing their relevance and utility. In this study, we introduce E-IUAutoMF, an extension of MF that incorporates explainable user-item based features. This model builds upon the Explainable Matrix Factorization (EMF) framework, utilizing contractive autoencoders to extract user and item features and providing explanations using neighborhood explanation techniques. The results of several experiments demonstrate that our proposed model E-IUAutoMF outperforms the other baselines models.
It is hard to avoid recommender systems (RS) these days which play a vital role in various domains, such as e-commerce, online streaming platforms, and personalized content delivery. These systems assist users in disc...
详细信息
ISBN:
(纸本)9798400701245
It is hard to avoid recommender systems (RS) these days which play a vital role in various domains, such as e-commerce, online streaming platforms, and personalized content delivery. These systems assist users in discovering relevant items based on their preferences and past interactions. A variety of methods are developed by RS communities, to address the accuracy issue. But, most of these methods are sequential and omit the item features and user side information which contains relevant and rich information that can increase the accuracy of these systems. However, enhancing the accuracy of recommendations often comes at the expense of increased computational costs. This PhD thesis aims to address the challenge of improving the accuracy of RS following a hybrid approach that allows leveraging user-item features based on some relevant state-of-the-art models and using deep learning techniques such as the contractive autoencoder (CAE) while optimizing the cost of computation using parallel/distributed paradigms.
Single-cell sequencing technologies are widely used to discover the evolutionary relationships and the differences in cells. Since dropout events may frustrate the analysis, many imputation approaches for single-cell ...
详细信息
Single-cell sequencing technologies are widely used to discover the evolutionary relationships and the differences in cells. Since dropout events may frustrate the analysis, many imputation approaches for single-cell RNA-seq data have appeared in previous attempts. However, previous imputation attempts usually suffer from the over-smooth problem, which may bring limited improvement or negative effect for the downstream analysis of single-cell RNA-seq data. To solve this difficulty, we propose a novel two-stage diffusion-denoising method called SCDD for large-scale single-cell RNA-seq imputation in this paper. We introduce the diffusion i.e. a direct imputation strategy using the expression of similar cells for potential dropout sites, to perform the initial imputation at first. After the diffusion, a joint model integrated with graph convolutional neural network and contractive autoencoder is developed to generate superposition states of similar cells, from which we restore the original states and remove the noise introduced by the diffusion. The final experimental results indicate that SCDD could effectively suppress the over-smooth problem and remarkably improve the effect of single-cell RNA-seq downstream analysis, including clustering and trajectory analysis.
Harmonic mapping provides a natural way of mapping two manifolds by minimizing distortion induced by the mapping. However, most applications are limited to mapping between 2D and/or 3D spaces owing to the high computa...
详细信息
Harmonic mapping provides a natural way of mapping two manifolds by minimizing distortion induced by the mapping. However, most applications are limited to mapping between 2D and/or 3D spaces owing to the high computational cost. We propose a novel approach, the harmonic autoencoder (HAE), by approximating a harmonic mapping in a data-driven way. The HAE learns a mapping from an input domain to a target domain that minimizes distortion and requires only a small number of input-target reference pairs. The HAE can be applied to high-dimensional applications, such as human-to-robot hand pose mapping. Our method can map from the input to the target domain while minimizing distortion over the input samples, covering the target domain, and satisfying the reference pairs. This is achieved by extending an existing neural network method called the contractive autoencoder. Starting from a contractive autoencoder, the HAE takes into account a distance function between point clouds within the input and target domains, in addition to a penalty for estimation error on reference points. For efficiently selecting a set of input-target reference pairs during the training process, we introduce an adaptive optimization criterion. We demonstrate that pairs selected in this way yield a higher-performance mapping than pairs selected randomly, and the mapping is comparable to that from pairs selected heuristically by the experimenter. Our experimental results with synthetic data and human-to-robot hand pose data demonstrate that our method can learn an effective mapping between the input and target domains.
Accurate predictions of stock markets are important for investors and other stakeholders of the equity markets to formulate profitable investment strategies. The improved accuracy of a prediction model even with a sli...
详细信息
Accurate predictions of stock markets are important for investors and other stakeholders of the equity markets to formulate profitable investment strategies. The improved accuracy of a prediction model even with a slight margin can translate into considerable monetary returns. However, the stock markets' prediction is regarded as an intricate research problem for the noise, complexity and volatility of the stocks' data. In recent years, the deep learning models have been successful in providing robust forecasts for sequential data. We propose a novel deep learning-based hybrid classification model by combining peephole LSTM with temporal attention layer (TAL) to accurately predict the direction of stock markets. The daily data of four world indices including those of U.S., U.K., China and India, from 2005 to 2022, are examined. We present a comprehensive evaluation with preliminary data analysis, feature extraction and hyperparameters' optimization for the problem of stock market prediction. TAL is introduced post peephole LSTM to select the relevant information with respect to time and enhance the performance of the proposed model. The prediction performance of the proposed model is compared with that of the benchmark models CNN, LSTM, SVM and RF using evaluation metrics of accuracy, precision, recall, F1-score, AUC-ROC, PR-AUC and MCC. The experimental results show the superior performance of our proposed model achieving better scores than the benchmark models for most evaluation metrics and for all datasets. The accuracy of the proposed model is 96% and 88% for U.K. and Chinese stock markets respectively and it is 85% for both U.S. and Indian markets. Hence, the stock markets of U.K. and China are found to be more predictable than those of U.S. and India. Significant findings of our work include that the attention layer enables peephole LSTM to better identify the long-term dependencies and temporal patterns in the stock markets' data. Profitable and timely tradi
Fault diagnosis of rotating machinery is vital to improve the security and reliability as well as avoid serious accidents. For instance, robust fault features are crucial to achieve a high diagnosis precision. However...
详细信息
Fault diagnosis of rotating machinery is vital to improve the security and reliability as well as avoid serious accidents. For instance, robust fault features are crucial to achieve a high diagnosis precision. However, traditional feature extraction methods rely on an abundant amount of expertise and human interference. As a breakthrough in fault diagnosis, deep learning holds the potential to automatically extract discriminative features without much prior knowledge and human interference. However, only a few deep learning models are designed to deal with noise and extract robust features. contractive autoencoder (CAE) is a potential tool to grasp the internal factors and directly obtain the hidden robust features by penalizing the Frobenius norm of the Jacobian matrix of the hidden features with respect to the inputs. Thus, this paper proposes a method based on stacked CAE for automatic robust features extraction and fault diagnosis of rotating machinery. Gearbox and bearing fault diagnosis experiments are conducted, and the testing accuracy of the proposed method is approximately 100% for both two cases and higher than that of other methods, which fully validates the effectiveness and superiority of the proposed method. In addition, experiments and correlation analysis under different signal-to-noise ratios (SNRs) are conducted. Results show that the diagnosis accuracies of the proposed method are higher than those of the stacked autoencoder (AE) network under each SNR, especially when under 0 dB, the testing accuracies of the proposed method are 4.14% and 5.88% higher than those of the stacked AE network in two case studies, and the correlation coefficients of the CAE are higher than those of the AE, which demonstrate the capability of CAE in mining more robust features compared to the regular AE automatically and the superiority of the proposed method in fault diagnosis.
The original hyperspectral data served as the initial features has the characteristics of high dimension and redundancy, which is not suitable for the subsequent analysis, so extracting feature information is needed. ...
详细信息
The original hyperspectral data served as the initial features has the characteristics of high dimension and redundancy, which is not suitable for the subsequent analysis, so extracting feature information is needed. The deep learning model has a strong ability in feature learning, but if the model has too many layers which will lead to the original information loss in the process of layer-by-layer feature learning and reduce the subsequent classification accuracy. To solve this problem, the paper proposed a deep learning model of hybrid structure with the contractive autoencoder and restricted boltzmann machine to extract the hyperspectral data feature information. First, through pre-processing the spectral data, the 2d spectrum data is converted into a one dimensional vector. Then, a hybrid model is constructed for unsupervised training and supervised learning for the hyperspectral data, and features are extracted from bottom to top gradually according to the hybrid model. Finally, the SVM classifier is adopted to enhance the classification ability of spectral data. The paper uses the hybrid model proposed to test for extracting features with two sets of AVIRIS data and compares with PCA and GCA methods. The experiment results show that the feature extraction algorithm based on hybrid depth model can get the better features, and have strong distinguish performance, and can get better classification accuracy by the SVM algorithm.
暂无评论