Search

Abstract With the development of brain-computer interface industry, large amounts of related applications have entered people’s vision. BCI applications based on steady-state visual evoked potentials (SSVEP) are widely used because they do not require pre-training and have high information transmission rates. However, in the actual use of SSVEP stimulus paradigm, the subjects will produce visual fatigue with the use, and fatigue will affect the transmission efficiency. In this experiment, an experimental environment consisting of two paradigm stimulus frequencies (7.5 Hz, 15 Hz), three resolutions (800 × 600, 1280 × 720, 1920 × 1080) and three refresh rates (120 Hz, 240 Hz, 360 Hz) is set up. The Likert scale is used to collect subjective fatigue and preference scores, and the EEG acquisition system and eye tracker are used to collect objective data. Using the proposed information entropy-CRITIC algorithm to combine subjective and objective indicators, a fatigue assessment system (display screen fitness-DSF) is innovated to score different experimental environments. The higher the DSF score, the better the visual experience. The results show that when using the 7.5 Hz SSVEP paradigm, the combination of 360 Hz and 1920 × 1080 can bring the best visual experience. When using the 15 Hz SSVEP paradigm, the combination of 240 Hz and 1280 × 720 is the best. DSF provides powerful help for hardware and software selection guidance and vision protection when using SSVEP-based BCI applications.

Abstract This paper presents a novel approach to the phase space reconstruction technique, fractional-order phase space reconstruction (FOSS), which generalizes the traditional integer-order derivative-based method. By leveraging fractional derivatives, FOSS offers a novel perspective for understanding complex time series, revealing unique properties not captured by conventional methods. We further develop the multi-span transition entropy component method (MTECM-FOSS), an advanced complexity measurement technique that builds upon FOSS. MTECM-FOSS decomposes complexity into intra-sample and inter-sample components, providing a more comprehensive understanding of the dynamics in multivariate data. In simulated data, we observe that lower fractional orders can effectively filter out random noise. Time series with diverse long- and short-term memory patterns exhibit distinct extremities at different fractional orders. In practical applications, MTECM-FOSS exhibits competitive or superior classification performance compared to state-of-the-art algorithms when using fewer features, indicating its potential for engineering tasks.

objective: Early diagnosis of infant cerebral palsy (CP) is very important for infant health. In this paper, we present a novel training-free method to quantify infant spontaneous movements for predicting CP. Methods: Unlike other classification methods, our method turns the assessment into a clustering task. First, the joints of the infant are extracted by the current pose estimation algorithm, and the skeleton sequence is segmented into multiple clips through a sliding window. Then we cluster the clips and quantify infant CP by the number of cluster classes. Results: The proposed method was tested on two datasets, and achieved state-of-the-arts (SOTAs) on both datasets using the same parameters. What’s more, our method is interpretable with visualized results. Conclusion: The proposed method can quantify abnormal brain development in infants effectively and be used in different datasets without training. Significance: Limited by small samples, we propose a training-free method for quantifying infant spontaneous movements. Unlike other binary classification methods, our work not only enables continuous quantification of infant brain development, but also provides interpretable conclusions by visualizing the results. The proposed spontaneous movement assessment method significantly advances SOTAs in automatically measuring infant health.

Summary: Combination therapy is a promising approach in treating multiple complex diseases. However, the large search space of available drug combinations exacerbates challenge for experimental screening. To predict synergistic drug combinations in different cancer cell lines, we propose an improved deep forest-based method, ForSyn, and design two forest types embedded in ForSyn. ForSyn handles imbalanced and high-dimensional data in medium-/small-scale datasets, which are inherent characteristics of drug combination datasets. Compared with 12 state-of-the-art methods, ForSyn ranks first on four metrics for eight datasets with different feature combinations. We conduct a systematic analysis to identify the most appropriate configuration parameters. We validate the predictive value of ForSyn with cell-based experiments on several previously unexplored drug combinations. Finally, a systematic analysis of feature importance is performed on the top contributing features extracted by ForSyn. The resulting key genes may play key roles on corresponding cancers. Motivation: Combination therapy has shown promise as a treatment for complex diseases such as cancer. Synergistic drug combinations can offer increased therapeutic efficacy and reduce toxicity compared with single drugs. However, class imbalances in datasets have complicated the use of computational tools, such as deep learning, for synergistic drug prediction. We propose an improved deep forest-based model, ForSyn, to address the above problem on imbalanced, medium- or small-scale datasets with high dimensionality.

Abstract Cross‐coupled microwave filter with flexible topology is an essential component in a modern communication system. Among the various filter networks, the use of an extended box section can effectively reduce the difficulty of filter design, but the synthesis approach for this popular structure is not well developed. This paper presents the synthesis method based on the folded coupling matrix to derive the extended box sections. At the same time, an eighth‐degree filter with extended sixth‐degree box sections is synthesized to demonstrate the correctness of the approach. Finally, the new approach's feature to derive advanced circuits such as peculiar quintets is discussed, indicating that it has certain universality.

Early detection and intervention of cerebral palsy can promote neural remodeling in the process of brain development, thus reducing the negative effects of cerebral palsy. In this paper, we proposed a novel method for early prediction of infant cerebral palsy based on General Movements Assessment (GMA) theory with RGB-D videos. Firstly, we explored the human pose recognition in supine position based on RGB-D videos. Then we further apply it to auto-GMA. Specifically, we employ current pose estimation method on RGB images to achieve the infant full body 2D key points. By combining the depth information, the 3D movement of infant in supine position can be obtained. Then the infant’s movement complexity index is achieved by extracting the infant’s whole-body movement characteristic. In order to verify the effectiveness of the method, we did some experiments on a public dataset consisting 12 real recorded infants’ movement RGB-D videos, with 4 of the samples were diagnosed as abnormal infants by a GMA expert. We use expert GMA ratings of these recorded movements as the gold standard. Our method achieved state-of-the-art with sensitivity of 100%, specificity of 87.5%, and accuracy of 91.7%. The results show that the method has great potential in assisting doctors in diagnose infant cerebral palsy.

Fault diagnosis is composed of two tasks, i.e., fault location detection and fault severity identification, which are both significant to equipment maintenance. The former can indicate where the defective component lies in, and the latter provides evidence on the residual life of the component. However, traditional fault diagnosis methods, like the time-based methods, frequency-based methods and time-frequency-based methods, can only achieve one goal every time. They are not able to produce highly representative features for dealing with above-mentioned two tasks simultaneously. In addition, there is a huge increase in the amount of monitoring data of equipment. There is urgent need for handling this massive data, obtaining highly discriminative features, and further producing accurate diagnosis results in the field of fault diagnosis. Aimed at these problems, a deep residual network based on multi-task learning is proposed, taking detection of fault location and judgment of fault severity into account simultaneously. This network is fed with two kinds of diagnostic information, which is helpful to mine the potential links between two tasks of fault diagnosis and generate very representative features. Moreover, based on maximizing activation value, a visualization method of role of deep neural network is proposed. It can break in the traditional way of using deep neural network as black box. A real bearing experiment validates that the proposed method is reliable and effective in bearing fault diagnosis.

In real industrial scenarios, intelligent fault diagnosis based on data-driven methods has been widely researched in the past decade. However, data scarcity is widespread in fault diagnosis tasks owning to the difficulties in collecting adequate data. As a result, there is an increasing demand for both researchers and engineers for fault identification with scarce data. To address this issue, an innovative domain-adaptive prototype-recalibrated network (DAPRN) based on a transductive learning paradigm and prototype recalibration strategy (PRS) is proposed, which has the potential to promote the generalization ability from the source domain to target domain in a few-shot fault diagnosis. Within this scheme, the DAPRN is composed of a feature extractor, a domain discriminator, and a label predictor. Concretely, the feature extractor is jointly optimized by the minimization of few-shot classification loss and the maximization of domain-discriminative loss. The cosine similarity-based label predictor, which is promoted by the PRS, is exploited to avoid the bias of naïve prototypes in the metric space and recognize the health conditions of machinery in the meta-testing process. The efficacy and advantage of DAPRN are validated by extensive experiments on bearing and gearbox datasets compared with seven popular and well-established few-shot fault diagnosis methods. In practical application, the proposed DAPRN is expected to solve more challenging few-shot fault diagnosis scenarios and facilitate practical fault identification problems in modern manufacturing.

Many of current human pose estimation methods based on depth images require training stage. However, the training stage costs huge work on making samples. And many methods for human pose occlusion condition cannot work well. In this study, a novel approach to estimate human pose with a depth image called model‐based recursive matching (MRM) is introduced. A human skeleton model with customised parameters is created based on T‐pose to fit different body types. The authors use depth image and 3D point cloud corresponding to input. In contrast to previous work, the proposed method avoids training step and can give an accurate estimation in the case of the human occlusion condition. They demonstrate the method by comparing to the method Kinect offered by using random forest on 20 human poses. And the ground truth of coordinates of pose joint is made by the motion capture system. The result shows that the proposed method not only works well on the general human pose but also can deal with human occlusion better. And the authors’ method can be also applied to the disabled people and other creatures.

With the application of deep convolutional neural networks, the performance of computer vision tasks has been improved to a new level. The construction of a deeper and more complex network allows the face recognition algorithm to obtain a higher accuracy, However, the disadvantages of large computation and storage costs of neural networks limit the further popularization of the algorithm. To solve this problem, we have studied the unified and efficient neural network face recognition algorithm under the condition of a single camera; we propose that the complete face recognition process consists of four tasks: face detection, in vivo detection, keypoint detection, and face verification; combining the key algorithms of these four tasks, we propose a unified network model based on a deep separable convolutional structure—UFaceNet. The model uses multisource data to carry out multitask joint training and uses the keypoint detection results to aid the learning of other tasks. It further introduces the attention mechanism through feature level clipping and alignment to ensure the accuracy of the model, using the shared convolutional layer network among tasks to reduce model calculations amount and realize network acceleration. The learning goal of multi-tasking implicitly increases the amount of training data and different data distribution, making it easier to learn the characteristics with generalization. The experimental results show that the UFaceNet model is better than other models in terms of calculation amount and number of parameters with higher efficiency, and some potential areas to be used.

The steady-state visual evoked potential (SSVEP)-based brain–computer interface (BCI) usually has the advantages of high information transfer rate (ITR) and no need for training. However, low frequencies, such as the human stride motion frequency, cannot easily induce SSVEP. To solve this problem, a light spot humanoid motion paradigm modulated by the change of brightness was designed in this study. The characteristics of the brain response to the motion paradigm modulated by the change of brightness were analyzed for the first time. The results showed that the designed paradigm could induce not only the high flicker frequency but also the modulation frequencies between the change of brightness and the motion in the primary visual cortex. Thus, the stride motion frequency can be recognized through the modulation frequencies by using the designed paradigm. Also, in an online experiment, this paradigm was employed to control a lower limb robot to achieve same frequency stimulation, which meant that the visual stimulation frequency was the same as the motion frequency of the robot. Also, canonical correlation analysis (CCA) was used to distinguish three different stride motion frequencies. The average accuracies of the classification in three walking speeds using the designed paradigm with the same and different high frequencies reached 87 and 95% respectively. Furthermore, the angles of the knee joint of the robot were obtained to demonstrate the feasibility of the electroencephalograph (EEG)-driven robot with same stimulation.

Human action recognition using 3D pose data has gained a growing interest in the field of computer robotic interfaces and pattern recognition since the availability of hardware to capture human pose. In this paper, we propose a fast, simple, and powerful method of human action recognition based on human kinematic similarity. The key to this method is that the action descriptor consists of joints position, angular velocity and angular acceleration, which can meet the different individual sizes and eliminate the complex normalization. The angular parameters of joints within a short sliding time window (approximately 5 frames) around the current frame are used to express each pose frame of human action sequence. Moreover, three modified KNN (k-nearest-neighbors algorithm) classifiers are employed in our method: one for achieving the confidence of every frame in the training step, one for estimating the frame label of each descriptor, and one for classifying actions. Additional estimating of the frame's time label makes it possible to address single input frames. This approach can be used on difficult, unsegmented sequences. The proposed method is efficient and can be run in real time. The research shows that many public datasets are irregularly segmented, and a simple method is provided to regularize the datasets. The approach is tested on some challenging datasets such as MSR-Action3D, MSRDailyActivity3D, and UTD-MHAD. The results indicate our method achieves a higher accuracy.