Improve Feature Representation Research Articles

Multi-level progressive parallel attention guided salient object detection for RGB-D images

Detecting salient objects in RGB-D images attracts more and more attention in recent years. It benefits from the widespread use of depth sensors and can be applied in the comprehensive understanding of RGB-D images. Existing models focus on double-stream networks which transfer from color stream to depth stream, but depth stream with one channel information cannot learn the same feature as color stream with three channels information even if HHA representation is adopted. In our works, RGB-D four-channels input is chosen, and meanwhile, progressive parallel spatial and channel attention mechanisms are performed to improve feature representation. Spatial and channel attention can pay more attention on partial positions and channels in the image which show higher response to salient objects. Both attentive features are optimized by attentive feature from higher layer, respectively, and parallel fed into recurrent convolutional layer to generate side-output saliency maps guided by saliency map from higher layer. Last multi-level saliency maps are fused together from multi-scale perspective. Experiments on benchmark datasets demonstrate that parallel attention mechanism and progressive optimization operation play an important role in improving the accuracy of salient object detection, and our model outperforms state-of-the-art models in evaluation matrices.

G-DipC: An Improved Feature Representation Method for Short Sequences to Predict the Type of Cargo in Cell-Penetrating Peptides.

Cell-penetrating peptides (CPPs) are functional short peptides with high carrying capacity. CPP sequences with targeting functions for the highly efficient delivery of drugs to target cells. In this paper, which is focused on the prediction of the cargo category of CPPs, a biocomputational model is constructed to efficiently distinguish the category of cargo carried by CPPs as macromolecular carriers among the seven known deliverable cargo categories. Based on dipeptide composition (DipC), an improved feature representation method, general dipeptide composition (G-DipC) is proposed for short peptide sequences and can effectively increase the abundance of features represented. Then linear discriminant analysis (LDA) is applied to mine some important low-dimensional features of G-DipC and a predictive model is built with the XGBoost algorithm. Experimental results with five-fold cross validation show that G-DipC improves accuracy by 25 and 5 percent compared with amino acid composition (AAC) and DipC, respectively. G-DipC is even found to be better than tripeptide composition (TipC). Thus, the proposed model provides a novel resource for the study of cell-penetrating peptides, and the improved dipeptide composition G-DipC can be widely adapted to determine the feature representation of other biological sequences.

Deep ensemble network based on multi-path fusion

Deep convolutional network is commonly stacked by vast number of nonlinear convolutional layers. Deep fused network can improve the training process of deep convolutional network due to its capability of learning multi-scale representations and of optimizing information flow. However, the depth in a deep fused network does not contribute to the overall performance significantly. Therefore, a deep ensemble network consisting of deep fused network and branch channel is proposed. First, two base networks are combined in a concatenation and fusion manner to generate a deep fused network architecture. Then, an ensemble block with embedded learning mechanisms is formed to improve feature representation power of the model. Finally, the computational efficiency is improved by introducing group convolution without loss of performance. Experimental results on the standard recognition tasks have shown that the proposed network achieves better classification performance and has superior generalization ability compared to the original residual networks.

End-to-End Partial Discharge Detection in Power Cables via Time-Domain Convolutional Neural Networks

The analysis of partial discharge (PD) signals has been identified as a standardized diagnostic tool in monitoring the condition of different electrical apparatus nowadays. In this paper, we propose a novel data-driven approach to detect PD pulses in power cables using one-dimensional convolutional neural networks (CNNs), a successful deep neural network approach. Applying this deep learning method, an end-to-end framework has been proposed considering the propagations of PD signals and noises in power cables. The proposed method uses PD pulses as input, automatically extracts meaningful features for waveforms of PD pulses, and finally detects PD. Most of the existing methods, which use traditional classifiers, such as support vector machines (SVMs) and multi-layer perceptron (MLP) have mainly focused on improving feature representation and extraction manually for this task. However, the proposed CNN-based detection algorithm captures important latent features for waveforms of PD pulses with the help of its automatic feature extraction capability from raw inputs. Our experimental results show that the proposed method is better than conventional SVMs and achieves 97.38% and 93.23% detection accuracies in end-to-end settings based on our theoretical model-generated and empirical real-world PD signals, respectively.

Semi Supervised Learning with Deep Embedded Clustering for Image Classification and Segmentation.

Deep neural networks usually require large labeled datasets to construct accurate models; however, in many real-world scenarios, such as medical image segmentation, labelling data is a time-consuming and costly human (expert) intelligent task. Semi-supervised methods leverage this issue by making use of a small labeled dataset and a larger set of unlabeled data. In this article, we present a flexible framework for semi-supervised learning that combines the power of supervised methods that learn feature representations using state-of-the-art deep convolutional neural networks with the deep embedded clustering algorithm that assigns data points to clusters based on their probability distributions and feature representations learned by the networks. Our proposed semi-supervised learning algorithm based on deep embedded clustering (SSLDEC) learns feature representations via iterations by alternatively using labeled and unlabeled data points and computing target distributions from predictions. During this iterative procedure the algorithm uses labeled samples to keep the model consistent and tuned with labeling, as it simultaneously learns to improve feature representation and predictions. SSLDEC requires few hyper-parameters and thus does not need large labeled validation sets, which addresses one of the main limitations of many semi-supervised learning algorithms. It is also flexible and can be used with many state-of-the-art deep neural network configurations for image classification and segmentation tasks. To this end, we implemented and tested our approach on benchmark image classification tasks as well as in a challenging medical image segmentation scenario. In benchmark classification tasks, SSLDEC outperformed several state-of-the-art semi-supervised learning methods, achieving 0.46% error on MNIST with 1000 labeled points, and 4.43% error on SVHN with 500 labeled points. In the iso-intense infant brain MRI tissue segmentation task, we implemented SSLDEC on a 3D densely connected fully convolutional neural network where we achieved significant improvement over supervised-only training as well as a semi-supervised method based on pseudo-labelling. Our results show that SSLDEC can be effectively used to reduce the need for costly expert annotations, enhancing applications such as automatic medical image segmentation.

Facial Landmark Detection via Attention-Adaptive Deep Network

Facial landmark detection is a key component of the face recognition pipeline as well as facial attribute analysis and face verification. Recently convolutional neural network-based face alignment methods have achieved significant improvement, but occlusion is still a major source of a hurdle to achieve good accuracy. In this paper, we introduce the attentioned distillation module in our previous work Occlusion-adaptive Deep Network (ODN) model, to improve performance. In this model, the occlusion probability of each position in high-level features are inferred by a distillation module. It can be learnt automatically in the process of estimating the relationship between facial appearance and facial shape. The occlusion probability serves as the adaptive weight on high-level features to reduce the impact of occlusion and obtain clean feature representation. Nevertheless, the clean feature representation cannot represent the holistic face due to the missing semantic features. To obtain exhaustive and complete feature representation, it is vital that we leverage a low-rank learning module to recover lost features. Considering that facial geometric characteristics are conducive to the low-rank module to recover lost features, the role of the geometry-aware module is, to excavate geometric relationships between different facial components. The role of attentioned distillation module is, to get rich feature representation and model occlusion. To improve feature representation, we used channel-wise attention and spatial attention. Experimental results show that our method performs better than existing methods.

Fast and robust dynamic hand gesture recognition via key frames extraction and feature fusion

Gesture recognition is a hot topic in computer vision and pattern recognition, which plays a vitally important role in natural human-computer interface. Although great progress has been made recently, fast and robust hand gesture recognition remains an open problem, since the existing methods have not well balanced the performance and the efficiency simultaneously. To bridge it, this work combines image entropy and density clustering to exploit the key frames from hand gesture video for further feature extraction, which can improve the efficiency of recognition. Moreover, a feature fusion strategy is also proposed to further improve feature representation, which elevates the performance of recognition. To validate our approach in a “wild” environment, we also introduce two new datasets called HandGesture and Action3D datasets. Experiments consistently demonstrate that our strategy achieves competitive results on Northwestern University, Cambridge, HandGesture and Action3D hand gesture datasets. Our code and datasets will release at https://github.com/Ha0Tang/HandGestureRecognition.

Recurrent Neural Network for Partial Discharge Diagnosis in Gas-Insulated Switchgear

The analysis of partial discharge (PD) signals has been identified as a standard diagnostic tool for monitoring the condition of different electrical apparatuses. This study proposes an approach to detecting PD patterns in gas-insulated switchgear (GIS) using a long short-term memory (LSTM) recurrent neural network (RNN). The proposed method uses phase-resolved PD (PRPD) signals as input, extracts low-level features, and finally, classifies faults in GIS. In the proposed method, LSTM networks can learn temporal dependencies directly from PRPD signals. Most existing models use support vector machines (SVMs) and mainly focus on improving feature representation and extraction manually to analyze PRPD signals. However, the proposed model captures important temporal features with the help of its low-level feature extraction capability from raw inputs. It outperforms conventional SVMs and achieves 96.74% classification accuracy for PRPDs in GIS.

Identification of DNA-Binding Proteins Using Mixed Feature Representation Methods.

DNA-binding proteins play vital roles in cellular processes, such as DNA packaging, replication, transcription, regulation, and other DNA-associated activities. The current main prediction method is based on machine learning, and its accuracy mainly depends on the features extraction method. Therefore, using an efficient feature representation method is important to enhance the classification accuracy. However, existing feature representation methods cannot efficiently distinguish DNA-binding proteins from non-DNA-binding proteins. In this paper, a multi-feature representation method, which combines three feature representation methods, namely, K-Skip-N-Grams, Information theory, and Sequential and structural features (SSF), is used to represent the protein sequences and improve feature representation ability. In addition, the classifier is a support vector machine. The mixed-feature representation method is evaluated using 10-fold cross-validation and a test set. Feature vectors, which are obtained from a combination of three feature extractions, show the best performance in 10-fold cross-validation both under non-dimensional reduction and dimensional reduction by max-relevance-max-distance. Moreover, the reduced mixed feature method performs better than the non-reduced mixed feature technique. The feature vectors, which are a combination of SSF and K-Skip-N-Grams, show the best performance in the test set. Among these methods, mixed features exhibit superiority over the single features.

Fast Prediction of Protein Methylation Sites Using a Sequence-Based Feature Selection Technique.

Protein methylation, an important post-translational modification, plays crucial roles in many cellular processes. The accurate prediction of protein methylation sites is fundamentally important for revealing the molecular mechanisms undergoing methylation. In recent years, computational prediction based on machine learning algorithms has emerged as a powerful and robust approach for identifying methylation sites, and much progress has been made in predictive performance improvement. However, the predictive performance of existing methods is not satisfactory in terms of overall accuracy. Motivated by this, we propose a novel random-forest-based predictor called MePred-RF, integrating several discriminative sequence-based feature descriptors and improving feature representation capability using a powerful feature selection technique. Importantly, unlike other methods based on multiple, complex information inputs, our proposed MePred-RF is based on sequence information alone. Comparative studies on benchmark datasets via vigorous jackknife tests indicate that our proposed MePred-RF method remarkably outperforms other state-of-the-art predictors, leading by a 4.5 percent average in terms of overall accuracy. A user-friendly webserver that implements the proposed method has been established for researchers' convenience, and is now freely available for public use through http://server.malab.cn/MePred-RF. We anticipate our research tool to be useful for the large-scale prediction and analysis of protein methylation sites.

Rotation-and-scale-invariant airplane detection in high-resolution satellite images based on deep-Hough-forests

This paper proposes a rotation-and-scale-invariant method for detecting airplanes from high-resolution satellite images. To improve feature representation capability, a multi-layer feature generation model is created to produce high-order feature representations for local image patches through deep learning techniques. To effectively estimate airplane centroids, a Hough forest model is trained to learn mappings from high-order patch features to the probabilities of an airplane being present at specific locations. To handle airplanes with varying orientations, patch orientation is defined and integrated into the Hough forest to augment Hough voting. The scale invariance is achieved by using a set of scale factors embedded in the Hough forest. Quantitative evaluations on the images collected from Google Earth service show that the proposed method achieves a completeness, correctness, quality, and F1-measure of 0.968, 0.972, 0.942, and 0.970, respectively, in detecting airplanes with arbitrary orientations and sizes. Comparative studies also demonstrate that the proposed method outperforms the other three existing methods in accurately and completely detecting airplanes in high-resolution remotely sensed images.

Improving Feature Representation Based on a Neural Network for Author Profiling in Social Media Texts.

We introduce a lexical resource for preprocessing social media data. We show that a neural network-based feature representation is enhanced by using this resource. We conducted experiments on the PAN 2015 and PAN 2016 author profiling corpora and obtained better results when performing the data preprocessing using the developed lexical resource. The resource includes dictionaries of slang words, contractions, abbreviations, and emoticons commonly used in social media. Each of the dictionaries was built for the English, Spanish, Dutch, and Italian languages. The resource is freely available.

Learning with click graph for query intent classification

Topical query classification, as one step toward understanding users' search intent, is gaining increasing attention in information retrieval. Previous works on this subject primarily focused on enrichment of query features, for example, by augmenting queries with search engine results. In this work, we investigate a completely orthogonal approach—instead of improving feature representation, we aim at drastically increasing the amount of training data. To this end, we propose two semisupervised learning methods that exploit user click-through data. In one approach, we infer class memberships of unlabeled queries from those of labeled ones according to their proximities in a click graph; and then use these automatically labeled queries to train classifiers using query terms as features. In a second approach, click graph learning and query classifier training are conducted jointly with an integrated objective. Our methods are evaluated in two applications, product intent and job intent classification. In both cases, we expand the training data by over two orders of magnitude, leading to significant improvements in classification performance. An additional finding is that with a large amount of training data obtained in this fashion, a classifier based on simple query term features can outperform those using state-of-the-art, augmented features.

Greyscale image processing for industrial applications

The processing of images of industrial components viewed in relatively good and relatively cluttered environments is discussed and illustrated. It is shown that a hypothesis generation and verification procedure can be used to improve feature representation. Finally, the requirements for robust modelling schemes are discussed.