Estimation Task Research Articles

Enhancing Depression Detection through Multimodal Emotion Recognition and Wearable Technologies

Traditional methods for diagnosing and treating depression often lack the precision to accurately identify the diverse symptoms and parameters associated with the disorder. With advancements in artificial intelligence and key technologies, multimodal emotion recognition has emerged as a promising approach to enhance depression detection. This paper explores the integration of physiological data such as EEG (electroencephalogram), ECG (electrocardiogram), and GSR (galvanic skin response), along with data from wearable devices, to develop a highquality feature extraction method for emotion analysis. Multiple datasets, including the SEED dataset involving EEG recordings from 15 participants during emotional video viewing, were utilized to collect comprehensive physiological and psychological data. Feature extraction involved preprocessing steps like noise reduction and normalization, followed by the application of decisionlevel and representationlevel fusion algorithms. Heart rate variability and skin conductance response features were analyzed to capture emotional indicators. An attentionbased Convolutional Neural Network (CNN) was employed to model both local and global facial features, and the Weighted Sum Pooling and Projection (WSPP) method was used for depression representation extraction. Multimodal fusion techniques, particularly modellevel fusion, significantly enhanced the learning of internal multimodal interactions, leading to improved recognition rates in Automatic Depression Estimation tasks compared to previous algorithms. The applications of this approach extend to smart home integration, daily emotion monitoring, health warning systems, and interactive wearable technologies, highlighting its potential for widespread use in enhancing user experience and mental health interventions.

GDE-Pose: A Real-Time Adaptive Compression and Multi-Scale Dynamic Feature Fusion Approach for Pose Estimation

This paper introduces a novel lightweight pose estimation model, GDE-pose, which addresses the current trade-off between accuracy and computational efficiency in existing models. GDE-pose builds upon the baseline YOLO-pose model by incorporating Ghost Bottleneck, a Dynamic Feature Fusion Module (DFFM), and ECA Attention to achieve more effective feature representation and selection. The Ghost Bottleneck reduces computational complexity, DFFM enhances multi-scale feature fusion, and ECA Attention optimizes the selection of key features. GDE-pose improves pose estimation accuracy while preserving real-time performance. Experimental results demonstrate that GDE-pose achieves higher accuracy on the COCO dataset, with a substantial reduction in parameters, over 80% fewer FLOPs, and an increased inference speed of 31 FPS, underscoring its exceptional lightweight and real-time capabilities. Ablation studies confirm the independent contribution of each module to the model’s overall performance. GDE-pose’s design highlights its broad applicability in real-time pose estimation tasks.

Temporal mechanisms underlying visual processing bias in peri-hand space.

The immediate space surrounding the hands has often been termed the peri-hand space (PHS), and is characterized by a smaller reaction time (RT), better detection, and enhanced accuracy for stimuli presented in this space, relative to those stimuli presented beyond this space. Such behavioral changes have been explained in terms of a biased allocation of cognitive resources such as perception, attention, and memory, for the efficient processing of information presented in the PHS. However, in two experiments, the current study shows that these cognitive biases seem to have an underlying temporal basis. The first experiment requires participants to perform a temporal bisection task, whereas the second experiment requires them to perform a verbal estimation task when stimuli are presented either near the hands or relatively far. Results from both experiments give evidence for slowing down of temporal mechanisms in the PHS - reflected in the form of temporal dilation for stimuli presented in the PHS relative to those presented further away. The slowing down of time in the PHS seems crucial in giving sufficient temporal allowance for the allocation of cognitive resources to prioritize the processing of information in the PHS. The findings are in line with the early anticipatory mechanisms associated with the PHS and seem to be driven by the switch/gate mechanism, and not the pacemaker component of the attentional gate model of time perception. Thus, the current study tries to integrate the theories of time perception with the peripersonal space literature.

360 Layout Estimation via Orthogonal Planes Disentanglement and Multi-View Geometric Consistency Perception.

Existing panoramic layout estimation solutions tend to recover room boundaries from a vertically compressed sequence, yielding imprecise results as the compression process often muddles the semantics between various planes. Besides, these data-driven approaches impose an urgent demand for massive data annotations, which are laborious and time-consuming. For the first problem, we propose an orthogonal plane disentanglement network (termed DOPNet) to distinguish ambiguous semantics. DOPNet consists of three modules that are integrated to deliver distortion-free, semantics-clean, and detail-sharp disentangled representations, which benefit the subsequent layout recovery. For the second problem, we present an unsupervised adaptation technique tailored for horizon-depth and ratio representations. Concretely, we introduce an optimization strategy for decision-level layout analysis and a 1D cost volume construction method for feature-level multi-view aggregation, both of which are designed to fully exploit the geometric consistency across multiple perspectives. The optimizer provides a reliable set of pseudo-labels for network training, while the 1D cost volume enriches each view with comprehensive scene information derived from other perspectives. Extensive experiments demonstrate that our solution outperforms other SoTA models on both monocular layout estimation and multi-view layout estimation tasks.

Comprehensive assessment of nonuniform image quality: Application to imaging near metal.

Comprehensive assessment of image quality requires accounting for spatial variations in (i) intensity artifact, (ii) geometric distortion, (iii) signal-to-noise ratio (SNR), and (iv) spatial resolution, among other factors. This work presents an ensemble of methods to meet this need, from phantom design to image analysis, and applies it to the scenario of imaging near metal. A modular phantom design employing a gyroid lattice is developed to enable the co-registered volumetric quantitation of image quality near a metallic hip implant. A method for measuring spatial resolution by means of local point spread function (PSF) estimation is presented and the relative fitness of gyroid and cubic lattices is examined. Intensity artifact, geometric distortion, and SNR maps are also computed. Results are demonstrated with 2D-FSE and MAVRIC-SL scan protocols on a 3T MRI scanner. The spatial resolution method demonstrates a worst-case error of 0.17 pixels for measuring in-plane blurring up to 3 pixels (full width at half maximum). The gyroid outperforms a cubic lattice design for the local PSF estimation task. The phantom supports four configurations toggling the presence/absence of both metal and structure with good spatial correspondence for co-registered analysis of the four quality factors. The marginal scan time to evaluate one scan protocol amounts to five repetitions. The phantom design can be fabricated in 2 days at negligible material cost. The phantom and associated analysis methods can elucidate complex image quality trade-offs involving intensity artifact, geometric distortion, SNR, and spatial resolution. The ensemble of methods is suitable for benchmarking imaging performance near metal.

Operational Support Estimator Networks.

In this work, we propose a novel approach called Operational Support Estimator Networks (OSENs) for the support estimation task. Support Estimation (SE) is defined as finding the locations of non-zero elements in sparse signals. By its very nature, the mapping between the measurement and sparse signal is a non-linear operation. Traditional support estimators rely on computationally expensive iterative signal recovery techniques to achieve such non-linearity. Contrary to the convolutional layers, the proposed OSEN approach consists of operational layers that can learn such complex non-linearities without the need for deep networks. In this way, the performance of non-iterative support estimation is greatly improved. Moreover, the operational layers comprise so-called generative super neurons with non-local kernels. The kernel location for each neuron/feature map is optimized jointly for the SE task during training. We evaluate the OSENs in three different applications: i. support estimation from Compressive Sensing (CS) measurements, ii. representation-based classification, and iii. learning-aided CS reconstruction where the output of OSENs is used as prior knowledge to the CS algorithm for enhanced reconstruction. Experimental results show that the proposed approach achieves computational efficiency and outperforms competing methods, especially at low measurement rates by significant margins.

Learning-based state estimation in distribution systems with limited real-time measurements

The task of state estimation in active distribution systems faces a major challenge due to the integration of different measurements with multiple reporting rates. As a result, distribution systems are essentially unobservable in real time, indicating the existence of multiple states that result in identical values for the available measurements. Certain existing approaches utilize historical data to infer the relationship between real-time available measurements and the state. Other learning-based methods aim to estimate the measurements acquired with a delay, generating pseudo-measurements. Our paper presents a methodology that utilizes the outcome of an underdetermined state estimator of an unobservable network state estimator to exploit information on the joint probability distribution between real-time available measurements and delayed ones to generate new physics-informed interpretable features. Through numerical simulations conducted on two realistic distribution grids of different size with insufficient real-time measurements, the proposed procedure showcases superior performance compared to existing state forecasting approaches and those relying on inferred pseudo-measurements.

Exploration of Imbalanced Regression in state-of-health estimation of Lithium-ion batteries

The state of health (SOH) estimation for lithium-ion batteries based on deep learning (DL) has made great progress. However, due to different electrochemical compositions of lithium-ion batteries, different ways of conducting experiments and other factors, the degradation process of some batteries shows longer early degradation time and shorter later degradation time, resulting in a long-tailed distribution of degradation data. This leads to the problem of data imbalance in SOH estimation tasks, which affects the accuracy of SOH estimation. This article explores the long-tailed distribution phenomenon in the field of batteries and the corresponding imbalanced regression problem it brings to the estimation of battery SOH. In addition, a method for improving model performance is proposed. Specifically, we use a quadratic interpolation and standardization method to analyze the battery data to ensure the consistency of data features. By discretized analysis of continuous problems, the label distribution smoothing (LDS) method is applied to deep neural networks to analyze and solve this imbalanced regression problem. By convolution processing with the kernel function and label distribution, the weights corresponding to different labels are calculated, which improves the estimation accuracy. We conducted battery aging experiments and verified that the degradation data follows a long-tailed distribution. The effectiveness of the final method was validated on our experimental data and a publicly available dataset.

Relationship Between Cognitive Estimation, Executive Functions, and Theory of Mind in Patients With Prefrontal Cortex Damage.

Conflicting evidence has arisen from the few studies that have examined the role of the prefrontal cortex and executive control functions in theory of mind (ToM). Moreover, the involvement of other cognitive domains in the ability to infer mental states is still under debate. This study aims to examine, in addition to the potential contribution of executive functions, the role of cognitive estimation in ToM abilities, given that cognitive estimation processes are strongly associated with some aspects of executive control functions. The cognitive estimation task, along with a set of neuropsychological tasks assessing executive functions, was administered to 30 patients with prefrontal cortex damage and 30 control subjects matched by gender, age, and education level. Patients with prefrontal cortex damage were impaired in all measures of executive functions, cognitive estimation, and theory of mind compared with control subjects. Regression analysis showed a significant interaction between executive measures and cognitive estimation in predicting ToM performance for patients with prefrontal cortex damage. Additionally, voxel-based lesion analysis identified a partially common bilaterally distributed prefrontal network involved in all three domains, centered within the ventral and dorsomedial areas with extension to the dorsolateral prefrontal cortex. Our findings highlight that, apart from executive functions, cognitive estimation plays a crucial role in the ability to interpret others' cognitive and emotional states in both patients with prefrontal cortex damage and control subjects.

Benchmark Dataset Applied to Quantitative Precipitation Estimation and Forecasting of Qinghai

We collected short-duration heavy precipitation events around weather radar since 2020 and get radar base data of these events. Then, we extracted radar products like Composite Reflectivity (CR), echo top height (ET) and Vertically Integrated Liquid Water (VIL) from radar base data after quality control as the input of dataset, and merge the surface precipitation of 6 minutes with background field using multi-grid variation to generate precipitation label by 6 minutes. Thus, we constructed two datasets artificial intelligence application QpefQH for short-duration heavy precipitation. QpefQH dataset has a scale of more than 15,000 images available for radar echo extrapolation and quantitative precipitation estimation tasks based on deep learning. QpefQH dataset also includes multi-source data such as satellite, numerical model, sounding, lightning, land surface temperature, pressure, wind and other observation data. QpefQH can be shared as a benchmark dataset for artificial intelligence research on short-duration heavy precipitation in Qinghai Province.

Generative learning assisted state-of-health estimation for sustainable battery recycling with random retirement conditions

Rapid and accurate state of health (SOH) estimation of retired batteries is a crucial pretreatment for reuse and recycling. However, data-driven methods require exhaustive data curation under random SOH and state of charge (SOC) retirement conditions. Here, we show that the generative learning-assisted SOH estimation is promising in alleviating data scarcity and heterogeneity challenges, validated through a pulse injection dataset of 2700 retired lithium-ion battery samples, covering 3 cathode material types, 3 physical formats, 4 capacity designs, and 4 historical usages with 10 SOC levels. Using generated data, a regressor realizes accurate SOH estimations, with mean absolute percentage errors below 6% under unseen SOC. We predict that assuming uniform deployment of the proposed technique, this would save 4.9 billion USD in electricity costs and 35.8 billion kg CO2 emissions by mitigating data curation costs for a 2030 worldwide battery retirement scenario. This paper highlights exploiting limited data for exploring extended data space using generative methods, given data can be time-consuming, expensive, and polluting to retrieve for many estimation and predictive tasks.

On proving the robustness of algorithms for early fault-tolerant quantum computers

The hope of the quantum computing field is that quantum architectures are able to scale up and realize fault-tolerant quantum computing. Due to engineering challenges, such ''cheap'' error correction may be decades away. In the meantime, we anticipate an era of ''costly'' error correction, or early fault-tolerant quantum computing. Costly error correction might warrant settling for error-prone quantum computations. This motivates the development of quantum algorithms which are robust to some degree of error as well as methods to analyze their performance in the presence of error. Several such algorithms have recently been developed; what is missing is a methodology to analyze their robustness. To this end, we introduce a randomized algorithm for the task of phase estimation and give an analysis of its performance under two simple noise models. In both cases the analysis leads to a noise threshold, below which arbitrarily high accuracy can be achieved by increasing the number of samples used in the algorithm. As an application of this general analysis, we compute the maximum ratio of the largest circuit depth and the dephasing scale such that performance guarantees hold. We calculate that the randomized algorithm can succeed with arbitrarily high probability as long as the required circuit depth is less than 0.916 times the dephasing scale.

Artificial intelligence in the diagnosis of cerebrovascular diseases using magnetic resonance imaging: A scoping review

AbstractThe field of radiology is currently undergoing revolutionary changes owing to the increasing application of artificial intelligence (AI). This scoping review identifies and summarizes the technical methods and clinical applications of AI applied to magnetic resonance imaging of cerebrovascular diseases (CVDs). Preferred Reporting Items for Systematic reviews and Meta‐Analyses extension for Scoping Reviews was adopted and articles listed in PubMed and Cochrane databases from January 1, 2018 to December 31, 2023, were assessed. In total, 67 articles met the eligibility criteria. We obtained a general overview of the field, including lesion types, sample sizes, data sources, and databases and found that nearly half of the studies used multisequence magnetic resonance as the input. Both classical machine learning and deep learning were widely used. The evaluation metrics varied according to the five main algorithm tasks of classification, detection, segmentation, estimation, and generation. Cross‐validation was primarily used with only one third of the included studies using external validation. We also illustrate the key questions of the CVD research studies and grade the clinical utility of their AI solutions. Although most attention is devoted to improving the performance of AI models, this scoping review provides information on the availability of algorithms, reliability of external validations, and consistency of evaluation metrics and may facilitate improved clinical applicability and acceptance.

Transcranial Direct Current Stimulation over the Posterior Parietal Cortex Increases Nontarget Retrieval during Visual Working Memory.

Visual working memory (VWM) requires precise feature binding. Previous studies have revealed a close relationship between the posterior parietal cortex (PPC) and feature binding during VWM; this study further examined their causal relationship through three transcranial direct current stimulation (tDCS) experiments. In Experiment 1 (N = 57), participants underwent three sessions of tDCS separately, including PPC stimulation, occipital cortex stimulation, and sham stimulation, and completed delayed estimation tasks for orientations before and after stimulation. Results showed that tDCS over PPC selectively prolonged recall response time (RT) and increased the probability of nontarget responses (a.k.a. failure of feature binding, pNT). In Experiment 2 (N = 29), combining metacognition estimation, we further investigated whether the effects of PPC stimulation were attributed to misbinding (i.e., participants self-reported "remembered" in nontarget responses) or informed guessing trials (participants self-reported "forgotten" in nontarget responses). We replicated the main findings in Experiment 1 and observed greater tDCS effects of PPC on RT in informed guessing trials while there are comparable effects on pNT in these two types of trials. In Experiment 3 (N = 28), we then examined whether the tDCS effects over PPC specifically influenced the memory retrieval process by using a change detection task. We found that PPC stimulation did not influence the recognition RT or accuracy. Together, this study provided direct causal evidence supporting the specific involvement of PPC in feature binding during VWM retrieval, from both aspects of speed and response preference, expanding our understanding of the neural basis of feature binding in VWM.

Heart Rate Estimation Algorithm Integrating Long and Short-Term Temporal Features

Non-contact heart rate monitoring from facial videos utilizing remote photoplethysmography (rPPG) has gained significant traction in remote health monitoring. Given that rPPG captures the dynamic blood flow within the human body and constitutes a time-series signal characterized by periodic properties, this study introduced a three-dimensional convolutional neural network (3D CNN) designed to simultaneously address long-term periodic and short-term temporal characteristics for effective rPPG signal extraction. Firstly, differential operations are employed to preprocess video data, enhancing the face’s dynamic features. Secondly, building upon the 3D CNN framework, multi-scale dilated convolutions and self-attention mechanisms were integrated to enhance the model’s temporal modeling capabilities further. Finally, interpolation techniques are applied to refine the heart rate calculation methodology. The experiments conducted on the UBFC-rPPG dataset indicate that, compared with the existing optimal algorithm, the average absolute error (MAE) and the root mean square error (RMSE) realized significant enhancements of approximately 28% and 35%. Additionally, through comprehensive analyses such as cross-dataset experiments and complexity analyses, the validity and stability of the proposed algorithm in the task of heart rate estimation were manifested.

CGLight: An effective indoor illumination estimation method based on improved convmixer and GauGAN

Illumination consistency is a key factor for seamlessly integrating virtual objects with real scenes in augmented reality (AR) systems. High dynamic range (HDR) panoramic images are widely used to estimate scene lighting accurately. However, generating environment maps requires complex deep network architectures, which cannot operate on devices with limited memory space. To address this issue, this paper proposes CGLight, an effective illumination estimation method that predicts HDR panoramic environment maps from a single limited field-of-view (LFOV) image. We first design a CMAtten encoder to extract features from input images, which learns the spherical harmonic (SH) lighting representation with fewer model parameters. Guided by the lighting parameters, we train a generative adversarial network (GAN) to generate HDR environment maps. In addition, to enrich lighting details and reduce training time, we specifically introduce the color consistency loss and independent discriminator, considering the impact of color properties on the lighting estimation task while improving computational efficiency. Furthermore, the effectiveness of CGLight is verified by relighting virtual objects using the predicted environment maps, and the root mean square error and angular error are 0.0494 and 4.0607 in the gray diffuse sphere, respectively. Extensive experiments and analyses demonstrate that CGLight achieves a balance between indoor illumination estimation accuracy and resource efficiency, attaining higher accuracy with nearly 4 times fewer model parameters than the ViT-B16 model.

From Avatars to Agents: Self-Related Cues Through Embodiment and Personalization Affect Body Perception in Virtual Reality.

Our work investigates the influence of self-related cues in the design of virtual humans on body perception in virtual reality. In a $2\times 2$ mixed design, 64 participants faced photorealistic virtual humans either as a motion-synchronized embodied avatar or as an autonomous moving agent, appearing subsequently with a personalized and generic texture. Our results unveil that self-related cues through embodiment and personalization yield an individual and complemented increase in participants' sense of embodiment and self-identification towards the virtual human. Different body weight modification and estimation tasks further showed an impact of both factors on participants' body weight perception. Additional analyses revealed that the participant's body mass index predicted body weight estimations in all conditions and that participants' self-esteem and body shape concerns correlated with different body weight perception results. Hence, we have demonstrated the occurrence of double standards through induced self-related cues in virtual human perception, especially through embodiment.

Deep Variational Network Toward Blind Image Restoration.

Blind image restoration (IR) is a common yet challenging problem in computer vision. Classical model-based methods and recent deep learning (DL)-based methods represent two different methodologies for this problem, each with their own merits and drawbacks. In this paper, we propose a novel blind image restoration method, aiming to integrate both the advantages of them. Specifically, we construct a general Bayesian generative model for the blind IR, which explicitly depicts the degradation process. In this proposed model, a pixel-wise non-i.i.d. Gaussian distribution is employed to fit the image noise. It is with more flexibility than the simple i.i.d. Gaussian or Laplacian distributions as adopted in most of conventional methods, so as to handle more complicated noise types contained in the image degradation. To solve the model, we design a variational inference algorithm where all the expected posteriori distributions are parameterized as deep neural networks to increase their model capability. Notably, such an inference algorithm induces a unified framework to jointly deal with the tasks of degradation estimation and image restoration. Further, the degradation information estimated in the former task is utilized to guide the latter IR process. Experiments on two typical blind IR tasks, namely image denoising and super-resolution, demonstrate that the proposed method achieves superior performance over current state-of-the-arts.

Learn to Adapt for Self-Supervised Monocular Depth Estimation.

Monocular depth estimation is one of the fundamental tasks in environmental perception and has achieved tremendous progress by virtue of deep learning. However, the performance of trained models tends to degrade or deteriorate when employed on other new datasets due to the gap between different datasets. Though some methods utilize domain adaptation technologies to jointly train different domains and narrow the gap between them, the trained models cannot generalize to new domains that are not involved in training. To boost the transferability of self-supervised monocular depth estimation models and mitigate the issue of meta-overfitting, we train the model in the pipeline of meta-learning and propose an adversarial depth estimation task. We adopt model-agnostic meta-learning (MAML) to obtain universal initial parameters for further adaptation and train the network in an adversarial manner to extract domain-invariant representations for easing meta-overfitting. In addition, we propose a constraint to impose upon cross-task depth consistency to compel the depth estimation to be identical in different adversarial tasks, which improves the performance of our method and smoothens the training process. Experiments on four new datasets demonstrate that our method adapts quite fast to new domains. Our method trained after 0.5 epoch achieves comparable results with the state-of-the-art methods trained at least 20 epochs.

Pollution Source Detection With Low‐Cost Low‐Accuracy Sensors Through Coupling Forward Data Assimilation and Inverse Optimization

AbstractData uncertainty affects the accuracy of pollution source detection (PSD), particularly in the background of low‐cost water quality sensing and low‐accuracy data challenge. This study aims to develop a novel PSD method to use low‐accuracy sensor data, namely, the method of coupled forward data Assimilation and inverse Optimization in PSD (A&O‐PSD). This approach primarily employs filtering strategies to handle observation errors and extract hidden trend information during forward water quality data assimilation, and then optimal estimation of pollution source information through inverse optimization with enhanced trend information matching, avoiding the non‐Gaussian distribution challenge of pollution source information. Both real‐world pollution events and semi‐synthetic cases were used to evaluate the methodology and compare its performance with the traditional optimization approach (T‐PSD). The results indicated that T‐PSD is significantly affected by observational and parameter noise, engendering noticeable biases in PSD under the low‐accuracy sensor conditions. In contrast, the A&O‐PSD could accomplish the estimation task of PSD in real‐world pollution events, with improved robustness against noise interference. Furthermore, A&O‐PSD achieved an accuracy improvement of over 10% compared to T‐PSD in estimating pollution source locations within the typical noise distribution range of most low‐accuracy sensors currently available, making it possible to use low‐accuracy data that would otherwise be unusable in T‐PSD. Overall, the A&O‐PSD method, combined with low‐cost low‐accuracy water quality sensing, offers an effective solution for watershed environmental management.