Convolutional Neural Network Network Research Articles

A novel multi-scale CNN and Bi-LSTM arbitration dense network model for low-rate DDoS attack detection

Low-rate distributed denial of service attacks, as known as LDDoS attacks, pose the notorious security risks in cloud computing network. They overload the cloud servers and degrade network service quality with the stealthy strategy. Furthermore, this kind of small ratio and pulse-like abnormal traffic leads to a serious data scale problem. As a result, the existing models for detecting minority and adversary LDDoS attacks are insufficient in both detection accuracy and time consumption. This paper proposes a novel multi-scale Convolutional Neural Networks (CNN) and bidirectional Long-short Term Memory (bi-LSTM) arbitration dense network model (called MSCBL-ADN) for learning and detecting LDDoS attack behaviors under the condition of limited dataset and time consumption. The MSCBL-ADN incorporates CNN for preliminary spatial feature extraction and embedding-based bi-LSTM for time relationship extraction. And then, it employs arbitration network to re-weigh feature importance for higher accuracy. At last, it uses 2-block dense connection network to perform final classification. The experimental results conducted on popular ISCX-2016-SlowDos dataset have demonstrated that the proposed MSCBL-ADN model has a significant improvement with high detection accuracy and superior time performance over the state-of-the-art models.

Emergency sign language recognition from variant of convolutional neural network (CNN) and long short term memory (LSTM) models

Sign language is the primary communication tool used by the deaf community and people with speaking difficulties, especially during emergencies. Numerous deep learning models have been proposed to solve the sign language recognition problem. Recently. Bidirectional LSTM (BLSTM) has been proposed and used in replacement of Long Short-Term Memory (LSTM) as it may improve learning long-team dependencies as well as increase the accuracy of the model. However, there needs to be more comparison for the performance of LSTM and BLSTM in LRCN model architecture in sign language interpretation applications. Therefore, this study focused on the dense analysis of the LRCN model, including 1) training the CNN from scratch and 2) modeling with pre-trained CNN, VGG-19, and ResNet50. Other than that, the ConvLSTM model, a special variant of LSTM designed for video input, has also been modeled and compared with the LRCN in representing emergency sign language recognition. Within LRCN variants, the performance of a small CNN network was compared with pre-trained VGG-19 and ResNet50V2. A dataset of emergency Indian Sign Language with eight classes is used to train the models. The model with the best performance is the VGG-19 + LSTM model, with a testing accuracy of 96.39%. Small LRCN networks, which are 5 CNN subunits + LSTM and 4 CNN subunits + BLSTM, have 95.18% testing accuracy. This performance is on par with our best-proposed model, VGG + LSTM. By incorporating bidirectional LSTM (BLSTM) into deep learning models, the ability to understand long-term dependencies can be improved. This can enhance accuracy in reading sign language, leading to more effective communication during emergencies.

Convolutional neural network intelligent fault diagnosis method for rotating machinery based on discriminant correlation analysis multi-domain feature fusion strategy

Aiming at solving the problems of limited training data, single input information, and limited diagnostic accuracy under the influence of strong background noise in fault diagnosis of rotating machinery, this paper proposes a fault diagnosis method based on the combination of discriminant correlation analysis (DCA) and convolutional neural network (CNN). Firstly, the original vibration signal is divided into several segments in the time domain, and the training data is directly processed by one CNN branch to extract multi-scale time domain features. Simultaneously, the divided data is subjected to discrete wavelet transform (DWT), and processed by another branch of CNN to extract multi-scale time-frequency features. Then, the DCA feature fusion mechanism is adopted to fuse the two-domain features extracted in the parallel branches to improve the model’ detection ability. Finally, the fused features are input into the deep CNN for training and learning to extract new features and output the classification results. Through the experimental analysis of two different types of data, the results show that the proposed method can be used for fault diagnosis of rotating machinery effectively. Compared with the single CNN network, the proposed method combines the multi-domain multi-scale feature extraction module with the DCA feature fusion module to enrich the feature information extraction ability. At the same time, the network performance is improved to get higher fault classification accuracy higher.

Exploring optimal pathways for enterprise procurement management systems based on fast neural modeling and semantic segmentation

Corporate procurement management assumes a pivotal role within the contemporary business landscape, yet confronts an array of challenges as markets continue to evolve and globalize. Conventional procurement management systems frequently grapple with issues of inefficiency, resource depletion, and noncompliance, necessitating the exploration of innovative avenues for optimization. This paper delves into the realm of risk mitigation associated with collusion behavior in the administration of intelligent procurement systems, presenting a novel procurement collusion identification model founded on a convolutional neural network (CNN) with reinforcement learning techniques. This framework commences with the application of a CNN and Long Short-Term Memory (LSTM) network for in-depth feature analysis and initial identification of historical procurement data, subsequently leveraging reinforcement learning methodologies to enhance the model's autonomy and intelligence for the purpose of optimization. Throughout the experimental phase, diverse domains of procurement data were meticulously selected for analysis. The empirical findings unequivocally demonstrate the model's proficiency, with an average recognition accuracy of 95.1% across five publicly available datasets. This performance surpasses existing machine learning methodologies employed in contemporary research and common recognition networks, thereby offering a pioneering reference point for the intelligent administration and optimization of future procurement systems.

Air Traffic Controller Fatigue Detection by Applying a Dual-Stream Convolutional Neural Network to the Fusion of Radiotelephony and Facial Data

The role of air traffic controllers is to direct and manage highly dynamic flights. Their work requires both efficiency and accuracy. Previous studies have shown that fatigue in air traffic controllers can impair their work ability and even threaten flight safety, which makes it necessary to carry out research into how to optimally detect fatigue in controllers. Compared with single-modality fatigue detection methods, multi-modal detection methods can fully utilize the complementarity between diverse types of information. Considering the negative impacts of contact-based fatigue detection methods on the work performed by air traffic controllers, this paper proposes a novel AF dual-stream convolutional neural network (CNN) architecture that simultaneously extracts controller radio telephony fatigue features and facial fatigue features and performs two-class feature-fusion discrimination. This study designed two independent convolutional processes for facial images and radio telephony data and performed feature-level fusion of the extracted radio telephony and facial image features in the fully connected layer, with the fused features transmitted to the classifier for fatigue state discrimination. The experimental results show that the detection accuracy of radio telephony features under a single modality was 62.88%, the detection accuracy of facial images was 96.0%, and the detection accuracy of the proposed AF dual-stream CNN network architecture reached 98.03% and also converged faster. In summary, a dual-stream network architecture based on facial data and radio telephony data is proposed for fatigue detection that is faster and more accurate than the other methods assessed in this study.

Carbon trading price forecasting based on parameter optimization VMD and deep network CNN–LSTM model

To meet carbon peak and neutrality targets, accurate carbon trading price forecasting is very important for enterprises making emission reduction decisions. By fusing convolutional neural network (CNN) and long short-term memory network (LSTM), the CNN–LSTM model is constructed. After variational mode decomposition (VMD), several intrinsic mode functions (IMFs) components are obtained and input into the CNN–LSTM model, thus constructing the combined sooty tern optimization algorithm (STOA)–VMD–CNN–LSTM forecasting model. To test this model, the carbon trading prices of the carbon emission trading markets of Hubei, Guangdong and Shenzhen were forecast. The prediction performance of the STOA–VMD–CNN–LSTM model is compared with ARIMA, BP, CNN and LSTM benchmark models and models combining different decomposition technologies. The international carbon trading price (EUR and CER) is used for prediction. Compared with other methods, the developed model makes fewer errors and achieves superior performance. Several important implications are provided for investors and risk managers involved in carbon financial products.

Automated detection of fatal cerebral haemorrhage in postmortem CT data

During the last years, the detection of different causes of death based on postmortem imaging findings became more and more relevant. Especially postmortem computed tomography (PMCT) as a non-invasive, relatively cheap, and fast technique is progressively used as an important imaging tool for supporting autopsies. Additionally, previous works showed that deep learning applications yielded robust results for in vivo medical imaging interpretation. In this work, we propose a pipeline to identify fatal cerebral haemorrhage on three-dimensional PMCT data. We retrospectively selected 81 PMCT cases from the database of our institute, whereby 36 cases suffered from a fatal cerebral haemorrhage as confirmed by autopsy. The remaining 45 cases were considered as neurologically healthy. Based on these datasets, six machine learning classifiers (k-nearest neighbour, Gaussian naive Bayes, logistic regression, decision tree, linear discriminant analysis, and support vector machine) were executed and two deep learning models, namely a convolutional neural network (CNN) and a densely connected convolutional network (DenseNet), were trained. For all algorithms, 80% of the data was randomly selected for training and 20% for validation purposes and a five-fold cross-validation was executed. The best-performing classification algorithm for fatal cerebral haemorrhage was the artificial neural network CNN, which resulted in an accuracy of 0.94 for all folds. In the future, artificial neural network algorithms may be applied by forensic pathologists as a helpful computer-assisted diagnostics tool supporting PMCT-based evaluation of cause of death.

Data augmentation techniques for classification of music genre

Music genre classification, a crucial aspect in the field of audio processing, has been widely investigated using different machine learning methods. However, there is still a need to determine the most optimal application between Long Short-Term Memory (LSTM) networks and Convolutional Neural Networks (CNNs) in the field of music genre classification, and the augmentation and enhancement of the data utilized for these machine learning techniques is still an important topic to study. The objective of this research is to identify the most effective solution and to improve the accuracy of music categorization. This will be achieved through an in-depth comparison of various data augmentation techniques and neural networks. In this study, both CNN and LSTM networks are optimized and trained. The dataset has been enhanced by extracting Mel Frequency Cepstral Coefficients (MFCC) data from music recordings using three different sampling techniques. The thorough approach undertaken facilitates a detailed assessment of the effectiveness of each method in classifying genres. The experimental findings show that the combination of CNN and random sampling outperforms all other tested algorithms, resulting in a significant enhancement in genre classification accuracy. This research provides valuable findings that can inform future studies in the pursuit of more effective techniques for classifying music genres.

A Reinforcement Learning Approach to Guide Web Crawler to Explore Web Applications for Improving Code Coverage

Web crawlers are widely used to automatically explore and test web applications. However, navigating the pages of a web application can be difficult due to dynamic page generation. In particular, the inputs for the web form fields can affect the resulting pages and subsequent navigation. Therefore, choosing the inputs and the order of clicks on a web page is essential for an effective web crawler to achieve high code coverage. This paper proposes a set of actions to quickly fill in web form fields and uses reinforcement learning algorithms to train a convolutional neural network (CNN). The trained agent, named iRobot, can autonomously select actions to guide the web crawler to maximize code coverage. We experimentally compared different reinforcement learning algorithms, neural networks, and actions. The results show that our CNN network with the proposed actions performs better than other neural networks in terms of branch coverage using the Deep Q-learning (DQN) or proximal policy optimization (PPO) algorithm. Furthermore, compared to previous studies, iRobot can increase branch coverage by about 1.7% while reducing training time to 12.54%.

Interpretable convolutional neural network for analyzing precipitation in the pre-rainy season of South China

Abstract The research and application of convolutional neural networks (CNNs) on statistical downscaling have been hampered by the fact that deep learning is highly dependent on sample size and is considered to be a black-box model. Therefore, a CNN model with transfer learning (CNN-TL) is proposed to study the pre-rainy season precipitation of South China. First, an augmented monthly dataset is created by sliding a fixed-length window over the daily circulation field and precipitation data for the entire year. Next, a base CNN network is pretrained on the augmented dataset, and then the network parameters are tuned on the actual monthly dataset from South China. Finally, guided backpropagation is conducted to obtain the distribution regions of the key features and explain the net. The coefficient of determination (R2) and root mean square error (RMSE) show that the CNN-TL model has higher explanatory power and better fitting performance than the feature extraction-based random forest (RF). Compared with the base CNN, the transfer learning approach can improve the explanatory power of the model by 10.29% and reduce the average RMSE by 6.82%. In addition, the interpretation results of the model show that the critical regions are primarily South China and its surrounding areas, including the Indochinese Peninsula, the Bay of Bengal, and the South China Sea. Furthermore, the ablation experiments and composite analysis illustrate that these regions are very important.

Multi-objective compression for CNNs via evolutionary algorithm

Filter pruning has been successfully applied in convolutional neural networks (CNN) to reduce model redundancy and speed up inference. Current pruning algorithms require setting a target value for a pruned model, which is not conducive to the rapid implementation in practical application. Firstly, for a target model, it may be necessary to pay attention to multiple performances at the same time, but it is hard to keep the best trade-off among indicators. Secondly, it is time-consuming to find a suitable model by pruning in a trial-and-error way. Therefore, this paper attempts to combine the most concerning indicators into a search target, rather than producing a model that only contains some pre-defined constraint. Specifically, the multi-object differential evolution algorithm is introduced into the search process, which is used to obtain a set of sparse structures that have taken all useful indicators into account. Then select proper sub-network to retrain from search results according to actual requirements. Experiments on multiple advanced CNN networks show that our algorithm can compress different indicators while keeping comparable or better performance than prior works. On ImageNet, pruning based on ResNet50 reduces FLOPs by 60.4%, while the accuracy of top-1 and top-5 only loses 1.09% and 0.57%, respectively.

The Use of CNN Network in the Generation of Clear Line Contours of Thangka Art

Thangka has been a crucial aspect of Buddhism for thousands of years. It acts as a visualization of the myths and legends that shaped Buddhism and attracted many devotees to worship. Furthermore, Thangka also serves as a reflection of the cultural, political and social aspect of Tibetan society in history. However, the paintings, especially the outlines, became extremely fragile after years of harsh environments. I utilized Deep learning, specifically the Convolutional neural network (CNN) as the main method for processing data. After numerous training the training loss gradually approaches 0, which means that the model no longer needs additional training. Under the CNN network I used max pooling which discards trivial information, rectified linear (ReLU) activation function, residential Network (ResNet), and other methods in this research to achieve a clear outline of Thangka. Through episodes of convolution and ReLU function, the image generated becomes gradually more defined in layer 1. However, as the image undergoes continuous process, it gradually becomes more abstract. Thus, the best results could be found in layer 1. This study aims to use the results from the CNN network convolution to aid researchers studying Thangka by helping to distinguish the outlines of the image effectively.

MFCEN: A lightweight multi-scale feature cooperative enhancement network for single-image super-resolution

<p>In recent years, significant progress has been made in single-image super-resolution with the advancements of deep convolutional neural networks (CNNs) and transformer-based architectures. These two techniques have led the way in the field of super-resolution technology research. However, performance improvements often come at the cost of a substantial increase in the number of parameters, thereby limiting the practical applications of super-resolution methods. Existing lightweight super-resolution methods, which primarily focus on single-scale feature extraction, lead to the issue of missing multi-scale features. This results in incomplete feature acquisition and poor reconstruction of the image. In response to these challenges, this paper proposed a lightweight multi-scale feature cooperative enhancement network (MFCEN). The network consists of three parts: shallow feature extraction, deep feature extraction, and image reconstruction. In the deep feature extraction part, a novel integrated multi-level feature module was introduced. Compared to existing CNN and transformer hybrid super-resolution networks, MFCEN significantly reduced the number of parameters while maintaining performance. This improvement was particularly evident at a scale factor of 3. The network introduced a novel comprehensive integrated multi-level feature module, leveraging the strong local perceptual capabilities of CNNs and the superior global information processing of transformers. It was designed with depthwise separable convolutions for extracting local information and a block-scale and global feature extraction module based on vision transformers (ViTs). While extracting the three scales of features, a satisfiability attention mechanism with a feed-forward network that can control the information was used to keep the network lightweight. Experiments demonstrated that the proposed model surpasses the reconstruction performance of the 498K-parameter SPAN model with a mere 488K parameters. Extensive experiments on commonly used image super-resolution datasets further validated the effectiveness of the network.</p>

Self-attention transformer unit-based deep learning framework for skin lesions classification in smart healthcare

Rising mortality rates in recent years have elevated melanoma to the ranks of the world’s most lethal cancers. Dermoscopy images (DIs) have been used in smart healthcare applications to determine medical features using deep transfer learning (DTL). DI-related lesions are widespread, have local features, and are associated with uncertainty. There are three components to our bi-branch parallel model: (1) the Transformer module (TM), (2) the self-attention unit (SAU), and (3) a convolutional neural network (CNN). With CNN and TM able to extract local and global features, respectively, a novel model has been developed to fuse global and local features using cross-fusion to generate fine-grained features. Parallel systems between the branches are merged using a feature-fusion architecture, resulting in a pattern that identifies the characteristics of a variety of lesions. Moreover, this paper proposes an optimized and lightweight CNN architecture version (optResNet-18) that discriminates skin cancer lesions with high accuracy. To verify the proposed method, the procedure evaluated the accuracy for the ISIC-2019 and the PH2 datasets as 97.48 and 96.87%, respectively, a significant difference over traditional CNN networks (e.g., ResNet-50 and ResNet-101) and the TM. The proposed model outperforms state-of-the-art performance metrics such as AUC, F1-score, specificity, precision, and recall. The proposed method can also be used as a generalizable model to diagnose different lesions in DIs with smart healthcare applications by combining DTL and medical imaging. With the proposed e-Health platform, skin diseases can be detected in real-time, which is crucial to speedy and reliable diagnostics.

Content Based Image Retrieval System Using CNN based Deep Learning Models

Digital information search archives on the internet and other platforms have been an essential part of our daily lives. With the use of huge amounts of digital images, it is paramount to develop efficient solutions to retrieve images seamlessly. Due to inadequate textual description, the task of automated image retrieval is quite complicated. Therefore, retrieval of images by analysing their visual content is an exciting task and research-worthy. Text based Image Search (TIBR) is time consuming as initially, every image has to be linked with a keyword or text to be extracted on query search. In the proposed system, an efficient algorithm for Content Based Image Retrieval (CBIR) using pre-trained CNN-based Deep Learning models to extract deep features of an image has been developed, which significantly increases the performance of the image retrieval process. Convolutional Neural Networks (CNN) has evolved as an efficient deep learning solution for the CBIR systems that work in image recognition applications. The results of retrieval are evaluated in terms of precision performance parameters. Emphasis has been given to the comparison of the efficiency of the system using two different pre-trained CNN-based Deep Learning models based on VGG16 and ResNet-50 architecture. Transfer learning enables us to develop robust and efficient working systems using pre-trained models. In the proposed system, pre-trained models have been used to derive features using a deep learning CNN network trained for a large image classification dataset ImageNet. This approach outperforms many contemporary CBIR systems. As most CBIR systems depend on query images and as most users of such systems are non-professional, we have developed a user interface for the built prototype systems. The efficiency of the CBIR systems with pre-trained models VGG16, and ResNet-50 are compared for better image retrieval and reliability. The proposed system has addressed a few existing problems in CBIR system design and developed a suitable solution with less semantic gap and better efficiency in image retrieval.

Deep Learning Model for Quality Assessment of Urinary Bladder Ultrasound Images using Multi-scale and Higher-order Processing.

Autonomous Ultrasound Image Quality Assessment (US-IQA) is a promising tool to aid the interpretation by practicing sonographers and to enable the future robotization of ultrasound procedures. However, autonomous US-IQA has several challenges. Ultrasound images contain many spurious artifacts, such as noise due to handheld probe positioning, errors in the selection of probe parameters and patient respiration during the procedure. Further, these images are highly variable in appearance with respect to the individual patient's physiology. We propose to use a deep Convolutional Neural Network (CNN), USQNet, which utilizes a Multi-scale and Local-to-Global Second-order Pooling (MS-L2GSoP) classifier to conduct the sonographer-like assessment of image quality. This classifier first extracts features at multiple scales to encode the inter-patient anatomical variations, similar to a sonographer's understanding of anatomy. Then, it uses second-order pooling in the intermediate layers (local) and at the end of the network (global) to exploit the second-order statistical dependency of multi-scale structural and multi-region textural features. The L2GSoP will capture the higher-order relationships between different spatial locations and provide the seed for correlating local patches, much like a sonographer prioritizes regions across the image. We experimentally validated the USQNet for a new dataset of the human urinary bladder ultrasound images. The validation involved first with the subjective assessment by experienced radiologists' annotation, and then with state-of-the-art CNN networks for US-IQA and its ablated counterparts. The results demonstrate that USQNet achieves a remarkable accuracy of 92.4% and outperforms the SOTA models by 3 - 14% while requiring comparable computation time.

A Bi-Directionally Fused Boundary Aware Network for Skin Lesion Segmentation.

It is quite challenging to visually identify skin lesions with irregular shapes, blurred boundaries and large scale variances. Convolutional Neural Network (CNN) extracts more local features with abundant spatial information, while Transformer has the powerful ability to capture more global information but with insufficient spatial details. To overcome the difficulties in discriminating small or blurred skin lesions, we propose a Bi-directionally Fused Boundary Aware Network (BiFBA-Net). To utilize complementary features produced by CNNs and Transformers, we design a dual-encoding structure. Different from existing dual-encoders, our method designs a Bi-directional Attention Gate (Bi-AG) with two inputs and two outputs for crosswise feature fusion. Our Bi-AG accepts two kinds of features from CNN and Transformer encoders, and two attention gates are designed to generate two attention outputs that are sent back to the two encoders. Thus, we implement adequate exchanging of multi-scale information between CNN and Transformer encoders in a bi-directional and attention way. To perfectly restore feature maps, we propose a progressive decoding structure with boundary aware, containing three decoders with six supervised losses. The first decoder is a CNN network for producing more spatial details. The second one is a Partial Decoder (PD) for aggregating high-level features with more semantics. The last one is a Boundary Aware Decoder (BAD) proposed to progressively improve boundary accuracy. Our BAD uses residual structure and Reverse Attention (RA) at different scales to deeply mine structural and spatial details for refining lesion boundaries. Extensive experiments on public datasets show that our BiFBA-Net achieves higher segmentation accuracy, and has much better ability of boundary perceptions than compared methods. It also alleviates both over-segmentation of small lesions and under-segmentation of large ones.

Novel Comparative Study of Covid-19 Detection from X-ray and CT Scan Images Using CNN and MLP Neural Networks

The coronavirus has caused the deaths of millions of people and has endangered the entire healthcare system. In order to count positive cases and stop the disease from spreading, Rapid clinical results may prevent the COVID-19 from spreading and help medical professionals treat patients while working under challenging circumstances.. Normal disease diagnosis using a laboratory test requires equipment and takes some time with the use of X-ray and chest CT Scan images, artificial intelligence techniques are extensively used to categorize the COVID-19. In this study we present an automatic detection approach for COVID-19 infection based on Chest CT and X-ray images using a Multilayer Perceptron (MLP) Neurons Network and a Convolutional Neural Network (CNN). The two models are evaluated in two classes, COVID-19 and normal images, for detection by Chest X-ray images we obtained 95,7% accuracy using MLP model and 90% accuracy using CNN model. For detection by Chest CT image we obtained, 80,60 % accuracy using the MLP model and 88,49 % accuracy using the CNN. The experimental results indicate that the proposed approach can achieve high accuracy in detecting COVID-19 from X-ray images, demonstrating the potential of using machine learning techniques in medical diagnosis.

Retrieving freeze-thaw states using deep learning with remote sensing data in permafrost landscapes

The soil freeze–thaw (FT) cycle is a critical component of the terrestrial cryosphere and plays a significant role in hydrological, ecological, climatic, and biogeochemical processes within permafrost landscapes. The FT states can be monitored with in-situ field measurements, but these procedures are costly and limited to single chosen sites. Remote sensing data provides the opportunity to collect information repeatedly across extensive geographical areas. To explore a more effective way to monitor the FT states in the terrestrial cryosphere, in this study, we used microwave and optical remote sensing data and introduced the Deep Learning approach to simulate the soil FT states in the western part of Nunavik, Canada.Two networks, Multilayer Perceptron (MLP) and Convolutional Neural Network (CNN), were trained and tested with over 35,000 and approximately 54,000 randomly selected data samples, respectively. The trained CNN networks outperformed the MLP networks, achieving the highest testing accuracy of 95.67% and the highest validation accuracy of 87.28% based on ground truth data from 32 measurement stations from all seasons across the year. This study proposed the reference periods concept for convenient labeling in data preparation and tested different combinations of influence variables to achieve better transferability of the method for future studies. Our findings offer a more effective way to monitor FT states in the terrestrial cryosphere, offering valuable insights into the consequences of climate change on permafrost landscapes. Moreover, the suggested deep learning approach can be easily expanded when additional input sources are accessible. This expansion has the potential to further improve the model's performance for the FT retrieval.

Research on a Non-Intrusive Load Recognition Algorithm Based on High-Frequency Signal Decomposition with Improved VI Trajectory and Background Color Coding

Against the backdrop of the current Chinese national carbon peak and carbon neutrality policies, higher requirements have been put forward for the construction and upgrading of smart grids. Non-intrusive Load Monitoring (NILM) technology is a key technology for advanced measurement systems at the end of the power grid. This technology obtains detailed power information about the load without the need for traditional hardware deployment. The key step to solve this problem is load decomposition and identification. This study first utilized the Long Short-Term Memory Denoising Autoencoder (LSTM-DAE) to decompose the mixed current signal of a household busbar and obtain the current signals of the multiple independent loads that constituted the mixed current. Then, the obtained independent current signals were combined with the voltage signals to generate multicycle colored Voltage–Current (VI) trajectories, which were color-coded according to the background. These color-coded VI trajectories formed a feature library. When the Convolutional Neural Network (CNN) was used for load recognition, in light of the influence of the hyperparameters on the recognition results, the Bayesian Optimization Algorithm (BOA) was used for optimization, and the optimized CNN network was employed for VI trajectory recognition. Finally, the proposed method was validated using the PLAID dataset. The experimental results show that the proposed method exhibited better performance in load decomposition and identification than current methods.