Class-dependent Features Research Articles

Weakly supervised learning for multi-class medical image segmentation via feature decomposition

Weakly supervised learning with image-level labels, releasing deep learning from highly labor-intensive pixel-wise annotation, has gained great attention for medical image segmentation. However, existing weakly supervised methods are mainly designed for single-class segmentation while leaving multi-class medical image segmentation rarely-explored. Different from natural images, label symbiosis, together with location adjacency, are much more common in medical images, making it more challenging for multi-class segmentation. In this paper, we propose a novel weakly supervised learning method for multi-class medical image segmentation with image-level labels. In terms of the multi-class classification backbone, a multi-level classification network encoding multi-scale features is proposed to produce binary predictions, together with the corresponding CAMs, of each class separately. To address the above issues (i.e., label symbiosis and location adjacency), a feature decomposition module based on semantic affinity is first proposed to learn both class-independent and class-dependent features by maximizing the inter-class feature distance. Through a cross-guidance loss to jointly utilize the above features, label symbiosis is largely alleviated. In terms of location adjacency, a mutually exclusive loss is constructed to minimize the overlap among regions corresponding to different classes. Experimental results on three datasets demonstrate the superior performance of the proposed weakly-supervised framework for both single-class and multi-class medical image segmentation. We believe the analysis in this paper would shed new light on future work for multi-class medical image segmentation. The source code of this paper is publicly available at https://github.com/HustAlexander/MCWSS.

Daily Activities of Elder Adults Using Optimized Deep Learning Model in China

The health and happiness of seniors can be influenced by the design of public spaces and buildings. Planning for age-friendly communities requires taking into account the wide variety of daily activities and public facility consumption among older persons. Yet, conventional approaches fall short when it comes to providing comprehensive, objective measures of the actions of the elderly. This research proposes an attention-based recognition approach by decomposing neural network outputs into class-dependent features using dictionary learning to improve discrimination performance. This research aimed to develop an empirical typology of activity of daily living (ADL) and its connection to health in China's ageing population. A deep learning-based model was rummaged to categorise the contributors in the Chinese Longitudinal Healthy Longevity Study (CLHLS) into subgroups based on their abilities with ADL. To be more precise, the study trains a category-based recognition network (CRN) by superimposing a basic but effective specific recognition encoding (SRE) unit on top of convolutional layers. In order to encode attention maps for each class, the SRE module uses the feature maps produced by the Convolutional Neural Networks (CNN) to learn a class-specific vocabulary. Class-wise adaptive feature refining is achieved by multiplying these attention maps by the input features. In this case, a gazelle optimization algorithm (GOA) handles the hyper-parameter tuning process of the proposed model. This case study is the first of its kind in a dense metropolitan region, and it uses objective measurements to evaluate the activity spaces of the elderly. This research provides empirical data for the promotion of healthy ageing in urban areas.

A Near-Field Area Object Detection Method for Intelligent Vehicles Based on Multi-Sensor Information Fusion

In order to solve the difficulty for intelligent vehicles in detecting near-field targets, this paper proposes a near-field object detection method based on multi-sensor information fusion. Firstly, the F-CenterFusion method is proposed to fuse the information from LiDAR, millimeter wave (mmWave) radar, and camera to fully obtain target state information in the near-field area. Secondly, multi-attention modules are constructed in the image and point cloud feature extraction networks, respectively, to locate the targets’ class-dependent features and suppress the expression of useless information. Then, the dynamic connection mechanism is used to fuse image and point cloud information to enhance feature expression capabilities. The fusion results are input into the predictive inference head network to obtain target attributes, locations, and other data. This method is verified by the nuScenes dataset. Compared with the CenterFusion method using mmWave radar and camera fusion information, the NDS and mAP values of our method are improved by 5.1% and 10.9%, respectively, and the average accuracy score of multi-class detection is improved by 22.7%. The experimental results show that the proposed method can enable intelligent vehicles to realize near-field target detection with high accuracy and strong robustness.

Optimization of Road Surface Wetness Classification Using Feature Selection Algorithms and Sensor Fusion

Due to the correlation between friction reduction and the road-covering water film height, knowledge about the current wetness level is of relevance for drivers and autonomous systems. A promising approach for wetness quantification is based on capacitive transducer arrays that enable to detect water spray ejected by the tires. Even though previous studies on this approach have shown the feasibility of wetness classification, optimization opportunities exist. While these previous investigations were limited to one feature selection algorithm, we study various algorithms and demonstrate the potential for optimization. Besides an application-specific algorithm that is capable of determining class-dependent features resulting in a performance improvement of more than 0.03 for the considered evaluation metrics, sequential forward floating selection in particular yields the most significant performance increase of more than 0.06. In addition, prior studies were limited to a test track with constrained conditions. Thus, in order to study the transferability of the preceding results, we present investigations on new experimental data acquired on public roads. The unknown and varying environmental conditions on public roads as well as a larger wheel speed range and steering angle effects are shown to significantly decrease classifier performance. We demonstrate that fusing two transducer arrays of different wheel arch liners increases performance by around 0.02. Here, a considerable information benefit can be attributed to the different transducer positions with design-related advantages. Furthermore, we show that the fusion with additional sensor data available in the vehicle results in a further performance improvement of more than 0.02.

Feature Selection Methods Based on Symmetric Uncertainty Coefficients and Independent Classification Information

Feature selection is a critical step in the data preprocessing phase in the field of pattern recognition and machine learning. The core of feature selection is to analyze and quantify the relevance, irrelevance, and redundancy between features and class labels. While existing feature selection methods give multiple explanations for these relationships, they ignore the multi-value bias of class-independent features and the redundancy of class dependent features. Therefore, a feature selection method (Maximal independent classification information and minimal redundancy, MICIMR) is proposed in this paper. Firstly, the relevance and redundancy terms of class independent characteristics are calculated respectively based on the symmetric uncertainty coefficient. Secondly, it calculates the relevance and redundancy terms of class-dependent features according to the independent classification information criterion. Finally, the selection criteria for these two characteristics are combined. To verify the effectiveness of the MICIMR algorithm, five feature selection methods are compared with the MICIMR algorithm on fifteen real datasets. The experimental results demonstrate that the MICIMR algorithm outperforms the other feature selection algorithms in terms of redundancy rate as well as classification accuracy (Gmean_macro and F1_macro).

The Relative Contribution of High-Gamma Linguistic Processing Stages of Word Production, and Motor Imagery of Articulation in Class Separability of Covert Speech Tasks in EEG Data

Word production begins with high-Gamma automatic linguistic processing functions followed by speech motor planning and articulation. Phonetic properties are processed in both linguistic and motor stages of word production. Four phonetically dissimilar phonemic structures “BA”, “FO”, “LE”, and “RY” were chosen as covert speech tasks. Ten neurologically healthy volunteers with the age range of 21–33 participated in this experiment. Participants were asked to covertly speak a phonemic structure when they heard an auditory cue. EEG was recorded with 64 electrodes at 2048 samples/s. Initially, one-second trials were used, which contained linguistic and motor imagery activities. The four-class true positive rate was calculated. In the next stage, 312 ms trials were used to exclude covert articulation from analysis. By eliminating the covert articulation stage, the four-class grand average classification accuracy dropped from 96.4% to 94.5%. The most valuable features emerge after Auditory cue recognition (~100 ms post onset), and within the 70–128 Hz frequency range. The most significant identified brain regions were the Prefrontal Cortex (linked to stimulus driven executive control), Wernicke’s area (linked to Phonological code retrieval), the right IFG, and Broca’s area (linked to syllabification). Alpha and Beta band oscillations associated with motor imagery do not contain enough information to fully reflect the complexity of speech movements. Over 90% of the most class-dependent features were in the 30-128 Hz range, even during the covert articulation stage. As a result, compared to linguistic functions, the contribution of motor imagery of articulation in class separability of covert speech tasks from EEG data is negligible.

Adaptive multi-parent crossover GA for feature optimization in epileptic seizure identification

EEG signal analysis involves multi-frequency non-stationary brain waves from multiple channels. Segmenting these signals, extracting features to obtain the important properties of the signal and classification are key aspects of detecting epileptic seizures. Despite the introduction of several techniques, it is very challenging when multiple EEG channels are involved. When many channels exist, a spatial filter is required to eliminate noise and extract relevant information. This adds a new dimension of complexity to the frequency feature space. In order to stabilize the classifier of the channels, feature selection is very important. Furthermore, and to improve the performance of a classifier, more data is required from EEG channels for complex problems. The increase of such data poses some challenges as it becomes difficult to identify the subject dependent bands when the channels increase. Hence, an automated process is required for such identification.The proposed approach in this work tends to tackle the multiple EEG channels problem by segmenting the EEG signals in the frequency domain based on changing spikes rather than the traditional time based windowing approach. While to reduce the overall dimensionality and preserve the class-dependent features an optimization approach is used. This process of selecting an optimal feature subset is an optimization problem. Thus, we propose an adaptive multi-parent crossover Genetic Algorithm (GA) for optimizing the features used in classifying epileptic seizures. The GA-based approach is used to optimize the various features obtained. It encodes the temporal and spatial filter estimates and optimize the feature selection with respect to the classification error. The classification was done using a Support Vector Machine (SVM).The proposed technique was evaluated using the publicly available epileptic seizure data from the machine learning repository of the UCI center for machine learning and intelligent systems. The proposed approach outperforms other ones and achieved a high level of accuracy. These results, indicate the ability of a multi-parent crossover GA in optimizing the feature selection process in EEG classification.

Determination of steel quality based on discriminating textural feature selection

Image data can be acquired from a product surface in real time by image sensor systems in chemical plants. For quality determination based on these image datasets, effective texture classification methodology is essential to handle highly dimensional images and to extract quality-related information from these product surface images. Wavelet texture analysis is useful for reducing the dimension and extracting textural information from images. Although wavelet texture analysis extracts only textural characteristics from images, the extracted features still contain unnecessary information for classification. The texture analysis method can be improved by retaining only class-dependent features and removing common features. In previous works, best basis and local discriminant basis are the most popular techniques for selecting an important basis from the wavelet packet basis. However, feature selection based on wavelet texture analysis has been studied for texture classification. Because previous methods are designed for wavelet coefficients with features for analysis, their performance is poor with wavelet texture analysis. We propose a novel texture classification methodology for quality determination based on feature selection using wavelet texture analysis. The proposed methodology applies the sequential forward floating selection (SFFS) algorithm as a feature selection strategy to select discriminating wavelet signatures using wavelet texture analysis. The proposed methodology is validated through quality determination for industrial steel surfaces. The results show that the proposed method has fewer classification errors with fewer number of features than previous methods.

A General Wrapper Approach to Selection of Class-Dependent Features

In this paper, we argue that for a C-class classification problem, C 2-class classifiers, each of which discriminating one class from the other classes and having a characteristic input feature subset, should in general outperform, or at least match the performance of, a C-class classifier with one single input feature subset. For each class, we select a desirable feature subset, which leads to the lowest classification error rate for this class using a classifier for a given feature subset search algorithm. To fairly compare all models, we propose a weight method for the class-dependent classifier, i.e., assigning a weight to each model's output before the comparison is carried out. The method's performance is evaluated on two artificial data sets and several real-world benchmark data sets, with the support vector machine (SVM) as the classifier , and with the RELIEF, class separability, and minimal-redundancy-maximal-relevancy (mRMR) as attribute importance measures. Our results indicate that the class-dependent feature subsets found by our approach can effectively remove irrelevant or redundant features, while maintaining or improving (sometimes substantially ) the classification accuracy, in comparison with other feature selection methods.

Analysis of class separation and combination of class-dependent features for handwriting recognition

In this paper, we propose a new approach to combine multiple features in handwriting recognition based on two ideas: feature selection-based combination and class dependent features. A nonparametric method is used for feature evaluation, and the first part of this paper is devoted to the evaluation of features in terms of their class separation and recognition capabilities. In the second part, multiple feature vectors are combined to produce a new feature vector. Based on the fact that a feature has different discriminating powers for different classes, a new scheme of selecting and combining class-dependent features is proposed. In this scheme, a class is considered to have its own optimal feature vector for discriminating itself from the other classes. Using an architecture of modular neural networks as the classifier, a series of experiments were conducted on unconstrained handwritten numerals. The results indicate that the selected features are effective in separating pattern classes and the new feature vector derived from a combination of two types of such features further improves the recognition rate.