Individual Feature Extraction Research Articles

Enhancing Convolutional Neural Networks via separately trained kernels for digit recognition

This paper introduces a novel methodology for pre-training neural networks. Instead of the traditional approach of running a single classifier for all objects, the method used in the research constructs individual classifiers for each object to extract unique features. These classifiers, in essence, act as a series of binary classifiers, each pre-training a distinct set of convolutional kernels. These individually trained kernels are then combined and processed by a comprehensive classifier. The methodology leverages the power of individual feature extraction and collaborative processing to enhance the overall performance of the classifier. The results demonstrate that this innovative pre-training approach leads to a significant improvement in classification performance, yielding higher accuracy compared to conventional methods. It underscores the benefits of incorporating individual object feature extraction and combined processing in the pre-training phase of neural network-based classification tasks. The study opens new paths for improving pre-training methods in neural networks, with potential applications in various fields that require high-accuracy object recognition. Future work will delve deeper into the potential challenges associated with model complexity and overfitting, and will also include a more comprehensive evaluation using extensive training, cross-validation, and independent test sets. This will further validate the effectiveness of our proposed pre-training methodology.

m B C C f : Multilevel Breast Cancer Classification Framework Using Radiomic Features

Breast cancer characterization remains a significant and challenging issue in contemporary medicine. Accurately distinguishing between malignant and benign breast lesions is crucial for effective diagnosis and treatment. The anatomical structure of malignant breast ultrasound images is more chaotic than that of benign images due to disease pathologies. However, texture-based analysis alone often fails to identify the extent of chaoticness in malignant breast ultrasound images due to their vague appearance with normal echo patterns, leading to missed diagnoses and increased mortality rates. To address this issue, we proposed an angular feature-based multilevel breast cancer classification framework mBCCf that aims to improve the accuracy and efficiency of classification. The proposed framework mimics the radiologist interpretation procedure by identifying the chaoticness on the periphery of the breast lesion in a breast ultrasound image (level-1). If the lesion contains an acute angle in any part of the periphery, it can be characterized as malignant or otherwise benign. However, solely relying on level-1 analysis may result in misclassification, especially when benign lesions exhibit echo patterns that resemble malignant ones. To overcome this limitation and to make the proposed system highly sensitive, advanced texture-based analysis (using combined shape, texture, and angular features) is performed (level-2). Finally, the performance of the proposed system is evaluated using a cross-dataset (consisting of 1293 breast ultrasound images) and compared with the different individual feature extraction techniques. Encouragingly, our system demonstrated an accuracy of 96.99% for classifying malignant and benign tumors, which is also validated using statistical analysis. The implications of our research lie in its potential to significantly improve breast cancer diagnosis by providing a reliable, efficient, and sensitive tool for radiologists.

Group behavior recognition algorithm for basketball video based on hidden tag mining strategy under gamification teaching and learning

The rapid development of Artificial Intelligence has facilitated the development of multimedia teaching and learning in the field of education, as well as providing intelligent methods for gamified teaching methods in the field of basketball. Currently, research on recognizing basketball video behavior primarily focuses on group behavior recognition methods for key individuals. However, there is currently no group behavior recognition method that can effectively distinguish features that reflect different events for temporal interactive behaviors such as dunks and steals in videos, as well as situations where key players are obstructed. The paper suggests a group behavior identification algorithm for basketball game videos based on the hidden tag mining technique in order to perform intelligent analysis in the fields of supplemental training and tactical analysis of gamified instruction. And this algorithm is integrated with the fusion of individual features to create the basketball video's group behavior identification algorithm with hidden label mining approach. According to the study's findings, the distribution of motion pattern features is more agglomerated and regular when the merging threshold is 0.6 and the best hyperparameter is 3. After the study applied the crucial individual feature extraction module, the overall group behavior identification accuracy increased by 6.44 %. The group behavior recognition algorithm for basketball videos that incorporates individual features and hidden label mining techniques outperforms the most recent patio-temporal domain-based group behavior recognition algorithm by 15.56 % when compared to the accuracy of commonly used algorithms. In summary, the study proposes a basketball video group behavior recognition algorithm with better performance. This algorithm achieves group behavior recognition in basketball game videos, promotes auxiliary training for students in gamified teaching mode, and provides a practical and feasible approach for intelligent analysis of other sports events. It also provides technical support for coaches and referees.

Breast Multiparametric MRI for Prediction of Neoadjuvant Chemotherapy Response in Breast Cancer: The BMMR2 Challenge.

Purpose To describe the design, conduct, and results of the Breast Multiparametric MRI for prediction of neoadjuvant chemotherapy Response (BMMR2) challenge. Materials and Methods The BMMR2 computational challenge opened on May 28, 2021, and closed on December 21, 2021. The goal of the challenge was to identify image-based markers derived from multiparametric breast MRI, including diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) MRI, along with clinical data for predicting pathologic complete response (pCR) following neoadjuvant treatment. Data included 573 breast MRI studies from 191 women (mean age [±SD], 48.9 years ± 10.56) in the I-SPY 2/American College of Radiology Imaging Network (ACRIN) 6698 trial (ClinicalTrials.gov: NCT01042379). The challenge cohort was split into training (60%) and test (40%) sets, with teams blinded to test set pCR outcomes. Prediction performance was evaluated by area under the receiver operating characteristic curve (AUC) and compared with the benchmark established from the ACRIN 6698 primary analysis. Results Eight teams submitted final predictions. Entries from three teams had point estimators of AUC that were higher than the benchmark performance (AUC, 0.782 [95% CI: 0.670, 0.893], with AUCs of 0.803 [95% CI: 0.702, 0.904], 0.838 [95% CI: 0.748, 0.928], and 0.840 [95% CI: 0.748, 0.932]). A variety of approaches were used, ranging from extraction of individual features to deep learning and artificial intelligence methods, incorporating DCE and DWI alone or in combination. Conclusion The BMMR2 challenge identified several models with high predictive performance, which may further expand the value of multiparametric breast MRI as an early marker of treatment response. Clinical trial registration no. NCT01042379 Keywords: MRI, Breast, Tumor Response Supplemental material is available for this article. © RSNA, 2024.

Analysis of Emotional Deconstruction and the Role of Emotional Value for Learners in Animation Works Based on Digital Multimedia Technology

With the rapid development of artificial intelligence and digital media technology, modern animation technology has greatly improved the creative efficiency of creators through computer-generated graphics, electronic manual painting, and other means, and its number has also experienced explosive growth. The intelligent completion of emotional expression identification within animation works holds immense significance for both animation production learners and the creation of intelligent animation works. Consequently, emotion recognition has emerged as a focal point of research attention. This paper focuses on the analysis of emotional states in animation works. First, by analyzing the characteristics of emotional expression in animation, the model data foundation for using sound and video information is determined. Subsequently, we perform individual feature extraction for these two types of information using gated recurrent unit (GRU). Finally, we employ a multiattention mechanism to fuse the multimodal information derived from audio and video sources. The experimental outcomes demonstrate that the proposed method framework attains a recognition accuracy exceeding 90% for the three distinct emotional categories. Remarkably, the recognition rate for negative emotions reaches an impressive 94.7%, significantly surpassing the performance of single-modal approaches and other feature fusion methods. This research presents invaluable insights for the training of multimedia animation production professionals, empowering them to better grasp the nuances of emotion transfer within animation and, thereby, realize productions of elevated quality, which will greatly improve the market operational efficiency of animation industry.

Multitask Deep Learning-Based Pipeline for Gas Leakage Detection via E-Nose and Thermal Imaging Multimodal Fusion

Innovative engineering solutions that are efficient, quick, and simple to use are crucial given the rapid industrialization and technology breakthroughs in Industry 5.0. One of the areas receiving attention is the rise in gas leakage accidents at coal mines, chemical companies, and home appliances. To prevent harm to both the environment and human lives, rapid and automated detection and identification of the gas type is necessary. Most of the previous studies used a single mode of data to perform the detection process. However, instead of using a single source/mode, multimodal sensor fusion offers more accurate results. Furthermore, the majority used individual feature extraction approaches that extract either spatial or temporal information. This paper proposes a deep learning-based (DL) pipeline to combine multimodal data acquired via infrared (IR) thermal imaging and an array of seven metal oxide semiconductor (MOX) sensors forming an electronic nose (E-nose). The proposed pipeline is based on three convolutional neural networks (CNNs) models for feature extraction and bidirectional long-short memory (Bi-LSTM) for gas detection. Two multimodal data fusion approaches are used, including intermediate and multitask fusion. Discrete wavelet transform (DWT) is utilized in the intermediate fusion to combine the spatial features extracted from each CNN, providing spectral–temporal representation. In contrast, in multitask fusion, the discrete cosine transform (DCT) is used to merge all of the features obtained from the three CNNs trained with the multimodal data. The results show that the proposed fusion approach has boosted the gas detection performance reaching an accuracy of 98.47% and 99.25% for intermediate and multitask fusion, respectively. These results indicate that multitask fusion is superior to intermediate fusion. Therefore, the proposed system is capable of detecting gas leakage accurately and could be used in industrial applications.

Intra-Native Accent Shared Features for Improving Neural Network-Based Accent Classification and Accent Similarity Evaluation

Accent similarity evaluation and accent identification are complex and challenging tasks for various applications due to the existence of variant types of native and non-native languages in the world. The absence of existing studies for the non-native and native English accent similarity evaluation and the limitation of individual feature extraction techniques for accent classifications have led us to propose a new model termed the intra-native accent feature sharing based native accent identification (NAI) framework using an English accent archive speech dataset. The NAI network was employed for non-native English accent classification, native English accent classification, and identification of native and non-native English accents. Finally, the accent similarity of native and non-native English accents was evaluated based on a delicate NAI pre-trained model. Moreover, the proposed approach has a high role in training data augmentation to overcome the challenge of a huge amount of training datasets demands of deep learning. The ordinary individual voice feature extraction with data augmentation and regularization techniques was the baseline for our work. The proposed approach boosted the accuracy of the baseline method with an average accuracy value of 3.7% -7.5% on different vigorous deep learning algorithms. The Quade test method for the performance comparison gave a 0.01 significant level (p-value) that proved that the proposed approach performed better than the baseline significantly. The model makes the rank for non-native English accents based on their similarity to native English accents and the proximity rank is Mandarin, Italian, German, French, Amharic, and Hindi.

A new non-invasive tagging method for leopard coral grouper (Plectropomus leopardus) using deep convolutional neural networks with PDE-based image decomposition

External tagging methods can aid in the research of leopard coral grouper (Plectropomus leopardus) in terms of its spatio-temporal behavior at population and individual scales. However, due to the strong exclusion ability and the damage to the body wall of P. leopardus, the retention rate of traditional invasive tagging methods is low. To develop a non-invasive identification method for P. leopardus, we adopted a multiscale image processing method based on matched filters with Gaussian kernels and partial differential equation (PDE) multiscale hierarchical decomposition with the deep convolutional neural network (CNN) models VGG19 and ResNet50 to extract shape and texture image features of individuals. Then based on image features, we used three classifiers Random forest (RF), support vector machine (SVM), and multilayer perceptron (MLP)) for individual recognition on sequential images of P. leopardus captured for 50 days. The PDE, ResNet50 and MLP combination obtained a maximum accuracy of 0.985 ± 0.045 on the test set. For individual temporal tracking recognition, feature extraction and model training were performed using images taken in 1-20 days. The classifier could achieve an accuracy of 0.960 ± 0.049 on the test set consisting of images collected in the periods of 20-50 days. The results show that CNNs with the PDE decomposition can effectively and accurately identify P. leopardus.

Extraction of individual gait features in mobile phone accelerometer

An algorithm for extracting individual signs of a person’s gait according to the data of a mobile phone accelerometer is considered. The developed algorithm includes the stage of data segmentation, correlation selection of motion patterns, and data clustering procedures. By selecting noisy segments, the quality of data clustering and the extraction of individual features is improved. The results of the algorithm work allow us to refine the existing models of changes in human motor behavior in personalized medicine and security systems. The article also discusses the features of data segmentation and the influence of external factors on the measurement results.

Computer Aided Diagnostic System for Blood Cells in Smear Images Using Texture Features and Supervised Machine Learning

Identification and diagnosis of leukemia earlier is a contentious issue in therapeutic diagnostics for reducing the rate of death among people with Acute Lymphoblastic Leukemia (ALL). The investigation of White Blood Cells (WBCs) is essential for the detection of ALL-leukaemia cells, for which blood smear images were being used. This study created an intelligent framework for identifying healthy blood cells from leukemic blood cells in blood smear images. The framework combines the features extracted by Center Symmetric Local Binary Pattern (CSLBP), Gabor Wavelet Transform (GWT), and Local Gradient Increasing Pattern (LGIP), the data was then fed into machine learning classifiers including Decision Tree (DT), Ensemble, K-Nearest Neighbor (KNN), Naïve Bayes (NB), and Random Forest (RF)). As the training set, the ALL-IDB2 database was utilized to create a balanced database with 260 blood smear images. Consequently, to generate the optimum feature set, a recommended model was established by using numerous individual and combined feature extraction methodologies. The investigational consequences demonstrate that the developed feature fusion strategy surpassed previous existing techniques, with an overall accuracy of 97.49 ± 1.02% utilizing Ensemble classifier.

A Hybrid Feature Extraction Method for Nepali COVID-19-Related Tweets Classification

COVID-19 is one of the deadliest viruses, which has killed millions of people around the world to this date. The reason for peoples' death is not only linked to its infection but also to peoples' mental states and sentiments triggered by the fear of the virus. People's sentiments, which are predominantly available in the form of posts/tweets on social media, can be interpreted using two kinds of information: syntactical and semantic. Herein, we propose to analyze peoples' sentiment using both kinds of information (syntactical and semantic) on the COVID-19-related twitter dataset available in the Nepali language. For this, we, first, use two widely used text representation methods: TF-IDF and FastText and then combine them to achieve the hybrid features to capture the highly discriminating features. Second, we implement nine widely used machine learning classifiers (Logistic Regression, Support Vector Machine, Naive Bayes, K-Nearest Neighbor, Decision Trees, Random Forest, Extreme Tree classifier, AdaBoost, and Multilayer Perceptron), based on the three feature representation methods: TF-IDF, FastText, and Hybrid. To evaluate our methods, we use a publicly available Nepali-COVID-19 tweets dataset, NepCov19Tweets, which consists of Nepali tweets categorized into three classes (Positive, Negative, and Neutral). The evaluation results on the NepCOV19Tweets show that the hybrid feature extraction method not only outperforms the other two individual feature extraction methods while using nine different machine learning algorithms but also provides excellent performance when compared with the state-of-the-art methods.

Classification of Acute Lymphoblastic Leukemia through the Fusion of Local Descriptors

Leukemia is characterized by an abnormal proliferation of leukocytes in the bone marrow and blood, which is usually detected by pathologists using a microscope to examine a blood smear. Leukemia identification and diagnosis in advance are a trending topic in medical applications for decreasing the death toll of individuals with Acute Lymphoblastic Leukemia (ALL). It is critical to analyze the white blood cells for the identification of ALL for which the blood smear images are utilized. This paper discusses and presents a micro-pattern descriptor, called Local Directional Number Pattern along with Multi-scale Weber Local Descriptor for feature extraction task to determine cancerous and noncancerous blood cells. A balanced dataset with 260 blood smear images from the ALL-IDB2 dataset was used as training data. Consequently, a proposed model was constructed by applying different individual and combined feature extraction methods, and fed into the machine learning classifiers (Decision Tree, Ensemble, K-Nearest Neighbors, Naïve Bayes, and Random Forest) to determine cancerous and noncancerous blood cells. Experimental results indicate that the developed feature fusion technique assured a reasonable performance compared to other existing works with a testing average accuracy of 97.69 ± 1.83% using Ensemble classifier.

Computer Vision-Based Classification of Flow Regime and Vapor Quality in Vertical Two-Phase Flow.

This paper presents a method to classify flow regime and vapor quality in vertical two-phase (vapor-liquid) flow, using a video of the flow as the input; this represents the first high-performing and entirely camera image-based method for the classification of a vertical flow regime (which is effective across a wide range of regimes) and the first image-based tool for estimating vapor quality. The approach makes use of computer vision techniques and deep learning to train a convolutional neural network (CNN), which is used for individual frame classification and image feature extraction, and a deep long short-term memory (LSTM) network, used to capture temporal information present in a sequence of image feature sets and to make a final vapor quality or flow regime classification. This novel architecture for two-phase flow studies achieves accurate flow regime and vapor quality classifications in a practical application to two-phase CO2 flow in vertical tubes, based on offline data and an online prototype implementation, developed as a proof of concept for the use of these models within a feedback control loop. The use of automatically selected image features, produced by a CNN architecture in three distinct tasks comprising flow-image classification, flow-regime classification, and vapor quality prediction, confirms that these features are robust and useful, and offer a viable alternative to manually extracting image features for image-based flow studies. The successful application of the LSTM network reveals the significance of temporal information for image-based studies of two-phase flow.

A multibranch, multitarget neural network for rapid point-source inversion in a microseismic environment: examples from the Hengill Geothermal Field, Iceland

SUMMARYDespite advanced seismological techniques, automatic source characterization for microseismic earthquakes remains difficult and challenging since current inversion and modelling of high-frequency signals are complex and time consuming. For real-time applications such as induced seismicity monitoring, the application of standard methods is often not fast enough for true complete real-time information on seismic sources. In this paper, we present an alternative approach based on recent advances in deep learning for rapid source-parameter estimation of microseismic earthquakes. The seismic inversion is represented in compact form by two convolutional neural networks, with individual feature extraction, and a fully connected neural network, for feature aggregation, to simultaneously obtain full moment tensor and spatial location of microseismic sources. Specifically, a multibranch neural network algorithm is trained to encapsulate the information about the relationship between seismic waveforms and underlying point-source mechanisms and locations. The learning-based model allows rapid inversion (within a fraction of second) once input data are available. A key advantage of the algorithm is that it can be trained using synthetic seismic data only, so it is directly applicable to scenarios where there are insufficient real data for training. Moreover, we find that the method is robust with respect to perturbations such as observational noise and data incompleteness (missing stations). We apply the new approach on synthesized and example recorded small magnitude (M ≤ 1.6) earthquakes at the Hellisheiði geothermal field in the Hengill area, Iceland. For the examined events, the model achieves excellent performance and shows very good agreement with the inverted solutions determined through standard methodology. In this study, we seek to demonstrate that this approach is viable for microseismicity real-time estimation of source parameters and can be integrated into advanced decision-support tools for controlling induced seismicity.

Electronic radar signal recognition based on wavelet transform and convolution neural network

With the continuous use of various new radar systems and complex radar systems, the electromagnetic environment is extremely deteriorated. The traditional emitter recognition methods have been difficult to meet the requirements of recognition performance in the rapidly changing battlefield environment. To solve this problem, a radar electronic signal recognition algorithm based on wavelet transform and deep learning is proposed in this paper. Starting from the radar reconnaissance system, the causes of signal preprocessing are analyzed, and the methods of signal denoising, signal normalization, signal intra pulse modulation recognition, multipath signal detection and suppression are deeply studied. In particular, the denoising algorithm based on threshold wavelet transform is proposed, which significantly improves the reliability of the algorithm. Aiming at the individual feature extraction of emitter signal, the extraction methods of emitter signal time domain feature, frequency domain feature, fuzzy function slice feature and cyclic spectrum feature based on wavelet transform are studied and analyzed, which provides stable and reliable classification features for emitter signal recognition. According to the characteristics of radar emitter signal, an optimized convolution neural network is designed, and the feature fusion processing is carried out at the decision-making level, which greatly improves the recognition effect and enhances the robustness of the recognition system. Experiments on radar data show that the fusion recognition rate is higher than any single feature and has strong robustness. In addition, compared with the traditional SVM and elm networks, the CNN network proposed in this paper can extract detailed features more effectively and improve the recognition rate of electronic radar.

Zoometric data extraction from drone imagery: the Arabian oryx (Oryx leucoryx)

SummaryData extraction from unmanned aerial vehicle (UAV) imagery has proved effective in animal surveys and monitoring, but to date has scarcely been used for detailed population analysis and individual animal feature extraction. We assessed the zoometric and feature extraction of the Arabian oryx (Oryx leucoryx) using data acquired from a captive population for comparison with reintroduced populations monitored by UAVs. Highly accurate scaled and geo-rectified imagery derived from UAV surveys allowed precise morphometric measurements of the oryx. The scaled top-view imagery combined with baseline data from known sex, age, weight and pregnancy status of captive individuals were used to develop predictive models. A bracketed index developed from the predictive models showed high accuracy for classifying the age group ≤16 months, animals with a weight >80 kg and pregnancy. The pregnancy classification decision tree model performed with 91.7% accuracy. The polynomial weight predictive model performed well with relatively high accuracy when using the total top-view surface measurement. Photogrammetrically processed UAV-acquired imagery can yield valuable zoometric data, feature extraction and modelling; it is a tool with a practical application for field biologists that can assist in the decision-making process for species conservation management.

COVID-19 Detection from Chest X-ray Images Using Feature Fusion and Deep Learning.

Currently, COVID-19 is considered to be the most dangerous and deadly disease for the human body caused by the novel coronavirus. In December 2019, the coronavirus spread rapidly around the world, thought to be originated from Wuhan in China and is responsible for a large number of deaths. Earlier detection of the COVID-19 through accurate diagnosis, particularly for the cases with no obvious symptoms, may decrease the patient’s death rate. Chest X-ray images are primarily used for the diagnosis of this disease. This research has proposed a machine vision approach to detect COVID-19 from the chest X-ray images. The features extracted by the histogram-oriented gradient (HOG) and convolutional neural network (CNN) from X-ray images were fused to develop the classification model through training by CNN (VGGNet). Modified anisotropic diffusion filtering (MADF) technique was employed for better edge preservation and reduced noise from the images. A watershed segmentation algorithm was used in order to mark the significant fracture region in the input X-ray images. The testing stage considered generalized data for performance evaluation of the model. Cross-validation analysis revealed that a 5-fold strategy could successfully impair the overfitting problem. This proposed feature fusion using the deep learning technique assured a satisfactory performance in terms of identifying COVID-19 compared to the immediate, relevant works with a testing accuracy of 99.49%, specificity of 95.7% and sensitivity of 93.65%. When compared to other classification techniques, such as ANN, KNN, and SVM, the CNN technique used in this study showed better classification performance. K-fold cross-validation demonstrated that the proposed feature fusion technique (98.36%) provided higher accuracy than the individual feature extraction methods, such as HOG (87.34%) or CNN (93.64%).

Feature Extraction Techniques with Analysis of Confusing Words for Speech Recognition in the Hindi Language

The research work presents experimental work to build a speaker-independent connected word Hindi speech recognition system using different feature extraction techniques with comparative analysis of confusing words. Comparative analysis of confusing words is essential to understand the reason for the speech recognition errors. Based on the error analysis, different feature extraction techniques, classification techniques, acoustic models, and pronunciation dictionaries can be selected to improve the speech recognition system's performance. Earlier studies for Hindi speech recognition lack detailed comparative analysis of confusing words for different feature extractions methods. As speaker-independent systems are developed for all, comparative analysis of confusing words is also presented for all feature extraction techniques. Speaker independent system was proposed with five states monophone based hidden Markov model (HMM) using HMM-based tool kit HTK. A Self-created data set of Hindi speech corpus has been used in the experiment. Feature extraction techniques such as linear predictive coding cepstral coefficients (LPCCs), mel frequency cepstral coefficients (MFCCs), and perceptual linear prediction coefficients (PLPs) were applied using delta, double delta, and energy parameters to evaluate the performance of the proposed methodology. The system was assessed by using different feature extraction techniques for speaker-independent mode. Research findings reveal that PLP coefficients show the highest recognition score, while LPCCs got the lowest recognition scores.Investigations also reveal that both PLP and MFCC coefficients are better than LPCC in speech recognition. Comparative analysis of confusing words shows that PLPs and MFCCs show fewer confusions than LPCCs and exhibit mostly the same pattern in the confusion analysis. Research outcomes also reveal that substitution errors are a significant cause of low recognition. It was also found that some words were recognized with individual feature extraction techniques only. Confusion analysis of the words indicates that words which have nasals, liquid, and fricative sound in first place exhibit more confusions. The investigation could improve speech recognition by choosing an appropriate feature extraction method and mixing the various feature extraction methods. The research outcomes can also be utilized to build linguistic resources for improving speech recognition. The results show that the developed recognition framework achieved the highest recognition word accuracy of 76.68% with PLPs for the speaker-independent model. The proposed system was also compared with existing similar work available.

Deep Feature-Level Sensor Fusion Using Skip Connections for Real-Time Object Detection in Autonomous Driving

Object detection is an important perception task in autonomous driving and advanced driver assistance systems. The visible camera is widely used for perception, but its performance is limited by illumination and environmental variations. For robust vision-based perception, we propose a deep learning framework for effective sensor fusion of the visible camera with complementary sensors. A feature-level sensor fusion technique, using skip connection, is proposed for the sensor fusion of the visible camera with the millimeter-wave radar and the thermal camera. The two networks are called the RV-Net and the TV-Net, respectively. These networks have two input branches and one output branch. The input branches contain separate branches for the individual sensor feature extraction, which are then fused in the output perception branch using skip connections. The RVNet and the TVNet simultaneously perform sensor-specific feature extraction, feature-level fusion and object detection within an end-to-end framework. The proposed networks are validated with baseline algorithms on public datasets. The results obtained show that the feature-level sensor fusion is better than baseline early and late fusion frameworks.

Accurate prediction of multi-label protein subcellular localization through multi-view feature learning with RBRL classifier.

Multi-label proteins can participate in carrier transportation, enzyme catalysis, hormone regulation and other life activities. Meanwhile, they play a key role in the fields of biopharmaceuticals, gene and cell therapy. This article proposes a prediction method called Mps-mvRBRL to predict the subcellular localization (SCL) of multi-label protein. Firstly, pseudo position-specific scoring matrix, dipeptide composition, position specific scoring matrix-transition probability composition, gene ontology and pseudo amino acid composition algorithms are used to obtain numerical information from different views. Based on the contribution of five individual feature extraction methods, differential evolution is used for the first time to learn the weight of single feature, and then these original features use a weighted combination method to fuse multi-view information. Secondly, the fused high-dimensional features use a weighted linear discriminant analysis framework based on binary weight form to eliminate irrelevant information. Finally, the best feature vector is input into the joint ranking support vector machine and binary relevance with robust low-rank learning classifier to predict the SCL. After applying leave-one-out cross-validation, the overall actual accuracy (OAA) and overall location accuracy (OLA) of Mps-mvRBRL on the training set of Gram-positive bacteria are both 99.81%. The OAA on the test sets of plant, virus and Gram-negative bacteria datasets are 97.24%, 98.55% and 98.20%, respectively, and the OLA are 97.16%, 97.62% and 98.28%, respectively. The results show that the model achieves good prediction performance for predicting the SCL of multi-label protein.