Prior Knowledge Research Articles

A path planning method based on deep reinforcement learning for AUV in complex marine environment

The potential of autonomous underwater vehicle (AUV) on future applications is significant due to advancements in autonomy and intelligence. Path planning is a critical technology for AUV to perform operational missions in complex marine environments. To this end, this paper proposes a path planning method for AUV based on deep reinforcement learning. Initially, considering actual requirements, a complex marine environment model containing underwater terrain, sonobuoy detection, and ocean currents is established. Subsequently, the corresponding state space, action space, and reward function are formulated. Furthermore, to address the inherent limitations of existing deep reinforcement learning algorithms in terms of training efficiency, a mixed experience replay (MER) strategy is proposed. This strategy aims to enhance the efficiency of sample learning by integrating prior knowledge and exploration experience. Lastly, a novel HMER-SAC algorithm for AUV path planning is proposed by integrating the Soft Actor–Critic (SAC) algorithm with the hierarchical reinforcement learning strategy and the MER strategy. The results of the simulation and experiment demonstrate that the method is capable of efficiently planning executable paths in complex marine environments and exhibits superior training efficiency, stability, and performance.

Adaptive backstepping and sliding mode control of a quadrotor

This article proposes a simple robust adaptive control architecture with minimum parameter estimation laws for trajectory tracking of a quadrotor subjected to parametric variations and environmental disturbances. This simple control architecture aims to achieve highly accurate, robust, and fast trajectory tracking of the quadrotor in a short time with low computational cost. Firstly, the quadrotor model is divided into altitude, attitude, and position subsystems for which appropriate control methods are designed without prior knowledge of the upper bound of external disturbances. A simple adaptive fractional-order sliding mode control (AFSMC) is designed to enhance the tracking of the altitude subsystem and estimate the upper bound of the disturbances. Then, a simple adaptive backstepping control (ABC) is developed for the horizontal position to generate the required roll and pitch orientations. The adaptation laws not only estimate the upper bound of the disturbances but also adjust the controller gains thereby enhancing the robustness of the ABC. A nonsingular fast terminal sliding mode control (NFTSMC) is incorporated with a finite-time disturbance observer (FDO) to accurately suppress the disturbances, and follow the target rotation angles within a short finite-time. Simulation results showed that the compounded control structure ensures accurate, fast, and robust tracking. The AFSMC can achieve the desired altitude with a settling time of 0.31 s and root mean square error (RMSE) of 0.0454 m. The ABC can attain the target horizontal position coordinates with a settling time of (0.47 s, 0.47 s) and RMSE of (0.0370 m, 0.0518 s). The NFTSMC-FDO can achieve the desired attitude angles with a settling time of (0.11 s, 0.13 s, 0.52 s) and RMSE of (0.0098 rad, 0.0092 rad, 0.0935 rad). Performance comparisons with existing control methods in terms of settling time and RMSE demonstrated that the proposed control architecture is superior.

Human-robot collaborative decision method of hexapod robot based on prior knowledge and negotiation strategy

Against the unstructured terrain environment, human drivers often rely on remote operating patterns to accomplish tasks with robots. With the gradual improvement in robot intelligence, a key problem that urgently requires attention is how to bridge cognitive and perceptual differences between human and robot in collaborative decision-making processes. Particularly effective resolution of differences in decision outcomes is crucial. Therefore, this paper utilizes the decision logic process and prior knowledge model of human drivers to form a robot's decision framework, narrowing the cognitive differences between human and robot. Besides, embedding the human-robot-environment features that affect the decision-making process as prior knowledge into the robot's decision-making system can effectively reduce perceptual gaps. Under the same decision framework, a negotiation strategy is proposed for forming a unified decision outcome between human and robot by finding an optimal concession rate. This study utilizes a hexapod robot simulated driving platform to gather experimental data and develop prior knowledge for robots. It also leverages virtual reality equipment and visual augmentation technologies to establish a remote human-robot collaborative decision-making experimental system. The experimental results conducted in complex obstacle terrain demonstrate the effectiveness of adopting the negotiation strategy for human-robot collaborative decision-making. Compared with the individual human or robot decision, the human-robot collaborative decision method significantly enhances robot stability, reduces energy consumption, and minimizes collisions.

A multicenter study on deep learning for glioblastoma auto-segmentation with prior knowledge in multimodal imaging.

A precise radiotherapy plan is crucial to ensure accurate segmentation of glioblastomas (GBMs) for radiation therapy. However, the traditional manual segmentation process is labor-intensive and heavily reliant on the experience of radiation oncologists. In this retrospective study, a novel auto-segmentation method is proposed to address these problems. To assess the method's applicability across diverse scenarios, we conducted its development and evaluation using a cohort of 148 eligible patients drawn from four multicenter datasets and retrospective data collection including noncontrast CT, multisequence MRI scans, and corresponding medical records. All patients were diagnosed with histologically confirmed high-grade glioma (HGG). A deep learning-based method (PKMI-Net) for automatically segmenting gross tumor volume (GTV) and clinical target volumes (CTV1 and CTV2) of GBMs was proposed by leveraging prior knowledge from multimodal imaging. The proposed PKMI-Net demonstrated high accuracy in segmenting, respectively, GTV, CTV1, and CTV2 in an 11-patient test set, achieving Dice similarity coefficients (DSC) of 0.94, 0.95, and 0.92; 95% Hausdorff distances (HD95) of 2.07, 1.18, and 3.95 mm; average surface distances (ASD) of 0.69, 0.39, and 1.17 mm; and relative volume differences (RVD) of 5.50%, 9.68%, and 3.97%. Moreover, the vast majority of GTV, CTV1, and CTV2 produced by PKMI-Net are clinically acceptable and require no revision for clinical practice. In our multicenter evaluation, the PKMI-Net exhibited consistent and robust generalizability across the various datasets, demonstrating its effectiveness in automatically segmenting GBMs. The proposed method using prior knowledge in multimodal imaging can improve the contouring accuracy of GBMs, which holds the potential to improve the quality and efficiency of GBMs' radiotherapy.

DZ-SLAM: A SAM-based SLAM algorithm oriented to dynamic environments

Precise localization is a fundamental prerequisite for the effective operation of Simultaneous Localization and Mapping (SLAM) systems. Traditional visual SLAM is based on static environments and therefore performs poorly in dynamic environments. While numerous visual SLAM methods have been proposed to address dynamic environments, these approaches are typically based on certain prior knowledge. This paper introduces DZ-SLAM, a dynamic SLAM algorithm that does not require any prior knowledge, based on ORB-SLAM3, to handle unknown dynamic elements in the scene. This work first introduces the FastSAM to enable comprehensive image segmentation. It then proposes an adaptive threshold-based dense optical flow approach to identify dynamic elements within the environment. Finally, combining FastSAM with optical flow method and embedding it into the SLAM framework to eliminate dynamic objects and improve positioning accuracy in dynamic environments. The experiment shows that compared with the original ORB-SLAM3 algorithm, the algorithm proposed in this paper reduces the absolute trajectory error by up to 96%; Compared to the most advanced algorithms currently available, the absolute trajectory error of our algorithm can be reduced by up to 46%. In summary, the proposed dynamic object segmentation method without prior knowledge can significantly reduce the positioning error of SLAM algorithm in various dynamic environments.

Medical-Knowledge-Based Graph Neural Network for Medication Combination Prediction.

Medication combination prediction (MCP) can provide assistance for experts in the more thorough comprehension of complex mechanisms behind health and disease. Many recent studies focus on the patient representation from the historical medical records, but neglect the value of the medical knowledge, such as the prior knowledge and the medication knowledge. This article develops a medical-knowledge-based graph neural network (MK-GNN) model which incorporates the representation of patients and the medical knowledge into the neural network. More specifically, the features of patients are extracted from their medical records in different feature subspaces. Then these features are concatenated to obtain the feature representation of patients. The prior knowledge, which is calculated according to the mapping relationship between medications and diagnoses, provides heuristic medication features according to the diagnosis results. Such medication features can help the MK-GNN model learn optimal parameters. Moreover, the medication relationship in prescriptions is formulated as a drug network to integrate the medication knowledge into medication representation vectors. The results reveal the superior performance of the MK-GNN model compared with the state-of-the-art baselines on different evaluation metrics. The case study manifests the application potential of the MK-GNN model.

An exploratory Q-matrix estimation method based on sparse non-negative matrix factorization.

Cognitive diagnostic assessment (CDA) is widely used because it can provide refined diagnostic information. The Q-matrix is the basis of CDA, and can be specified by domain experts or by data-driven estimation methods based on observed response data. The data-driven Q-matrix estimation methods have become a research hotspot because of their objectivity, accuracy, and low calibration cost. However, most of the existing data-driven methods require known prior knowledge, such as initial Q-matrix, partial q-vector, or the number of attributes. Under the G-DINA model, we propose to estimate the number of attributes and Q-matrix elements simultaneously without any prior knowledge by the sparse non-negative matrix factorization (SNMF) method, which has the advantage of high scalability and universality. Simulation studies are carried out to investigate the performance of the SNMF. The results under a wide variety of simulation conditions indicate that the SNMF has good performance in the accuracy of attribute number and Q-matrix elements estimation. In addition, a set of real data is taken as an example to illustrate its application. Finally, we discuss the limitations of the current study and directions for future research.

Pharmacy Undergraduates’ Epistemic Cognition and Medication Beliefs as Predictors of Conceptual Learning and Academic Progress

The importance of studies showing the impact of students' epistemic beliefs on their conceptual learning and academic progress is increasing. This study investigated pharmacy students’ domain-specific epistemic cognition and topic-specific beliefs related to their level of conceptual prior knowledge, learning and study progress during the first academic year. Data were collected from 125 students using a pre-test/post-test design with a multiple-choice questionnaire, an open-ended case task and a measure of domain-specific epistemic cognition and topic-specific beliefs concerning medication. The results showed that students’ prior knowledge predicted their academic performance and more sophisticated epistemic cognition was related to better conceptual understanding, faster study progress, and fewer anti-medication beliefs. Anti-medication beliefs hindered participants’ success in the case task and were related to weaker study progress. Our study shows that epistemic cognition and topic-specific beliefs play a role in students’ performance, learning, and study progress

Deep indoor illumination estimation based on spherical gaussian representation with scene prior knowledge

High dynamic range illumination estimation from a single low dynamic range image is a critical task in the fields of computer vision, graphics and augmented reality. However, directly learning the full HDR environment map or parametric lighting information from a single image is extremely difficult and inaccurate. As a result, we propose a two-stage network approach for illumination estimation that integrates spherical gaussian (SG) representation with scene prior knowledge. In the first stage, a convolutional neural network is utilized to generate material and geometric information about the scene, which serves as prior knowledge for lighting prediction. In the second stage, we model indoor environment illumination using 128 SG functions with fixed center direction and bandwidth, allowing only the amplitude to vary. Subsequently, a Transformer-based lighting parameter regressor is employed to capture the complex relationship between the input images with scene prior information and its SG illumination. Additionally, we introduce a hybrid loss function, which combines a masked loss for high-frequency illumination with a rendering loss for improving the visual quality. By training and evaluating the lighting model on the created SG illumination dataset, the proposed method achieves competitive results in both quantitative metrics and visual quality, outperforming state-of-the-art methods.

Net-zero carbon emission oriented Bi-level optimal capacity planning of integrated energy system considering carbon capture and hydrogen facilities

The integrated energy system (IES) is considered an efficient energy provision paradigm to improve the utilization efficiency of renewable generation (RG), reduce energy cost and improve energy supply reliability. However, the multi-dimensional uncertainties and the hydrogen utilization have posed enormous challenges to the optimal planning of the IES. This paper proposes a net-zero carbon emission oriented bi-level optimal planning model for the IES with full consideration of the multiple operational uncertainties in the IES (RG, multiple demands). The upper level is an optimal capacity configuration model, and the economic costs, energy selling profits, the penalty for power abandonment and energy supply shortage are settled as the objectives. The lower level is an optimal energy dispatch model with the objective function of minimizing the total operating cost of the IES on typical days. An integrated energy system framework consisting of multiple energy flows is formulated in the presence of carbon capture system (CCS) and hydrogen-to-power (H2P) facilities. To fully characterize the uncertainties of RG and multiple demands, a data-driven scenario generation method is adopted to expand the operational scenarios, in which probability distribution assumptions and prior knowledge are not necessary. Finally, the effectiveness of the proposed solution is extensively demonstrated through numerical simulation, and a sensitivity analysis is carried out to assess the impact of changes in electricity and gas prices on the IES planning results.

SpaGRA: Graph augmentation facilitates domain identification for spatially resolved transcriptomics

Recent advances in spatially resolved transcriptomics (SRT) have provided new opportunities for characterizing spatial structures of various tissues. Graph-based geometric deep learning have gained widespread adoption for spatial domain identification tasks. Currently, most methods define adjacency relation between cells or spots by their spatial distance in SRT data, which overlooks key biological interactions like gene expression similarities, and leads to inaccuracies in spatial domain identification. To tackle this challenge, we propose a novel method, SpaGRA (https://github.com/sunxue-yy/SpaGRA), for automatic multi-relationship construction based on graph augmentation. SpaGRA uses spatial distance as prior knowledge and dynamically adjusts edge weights with multi-head graph attention networks (GATs). This helps SpaGRA to uncover diverse node relationships and enhance message passing in geometric contrastive learning. Additionally, SpaGRA uses these multi-view relationships to construct negative samples, addressing sampling bias posed by random selection. Experimental results show that SpaGRA demonstrates superior domain identification performance on multiple datasets generated from different protocols. Using SpaGRA, we analyzed the functional regions in the mouse hypothalamus, identified key genes related to heart development in mouse embryos, and observed cancer-associated fibroblasts enveloping cancer cells in the latest Visium HD data. Overall, SpaGRA can effectively characterize spatial structures across diverse SRT datasets.

Particle size by design: Standardizing measurements for complex topical drug product assessment

The physicochemical and biopharmaceutical properties of drug substances and dosage forms can be significantly influenced by particle size. However, the diversity of equivalent particle diameters generated by different methods poses a fundamental challenge in particle size analysis. This study aimed to develop an Analytical Quality by Design (AQbD) approach to accurately assess the particle size of a complex formulation – clobetasol propionate (CP) 0.5 mg/g cream – through automated microscopy (AM) and laser light diffraction (LD). Additionally, Raman spectroscopy was utilized to determine the chemical composition of the formulation particles. In the AQbD approach, prior knowledge was considered for the construction of the Ishikawa diagram and estimate failure mode and effects analysis (FMEA). The methods were developed following the ICH Q8-Q10, and ICH Q14 guidelines, and validated according to ICH Q2, ISO 13320:2020, and EP2.9.31./USP<429>. Results indicate that a trade-off between the techniques must be established for a particle size by design: while LD offers higher throughput and more precise values at the expense of peak resolution and broadening, AM has higher variability but more reliable information in terms of size and shape analysis. The validated methods successfully demonstrated the implementation of an AQbD method in the definition of particle size methods.

ChatDiff: A ChatGPT-based diffusion model for long-tailed classification

Long-tailed data distributions have been a major challenge for the practical application of deep learning. Information augmentation intends to expand the long-tailed data into uniform distribution, which provides a feasible way to mitigate the data starvation of underrepresented classes. However, most existing augmentation methods face two significant challenges: (1) limited diversity in generated samples, and (2) the adverse effect of generated negative samples on downstream classification performance. In this paper, we propose a novel information augmentation method, named ChatDiff, to provide diverse positive samples for underrepresented classes, and eliminate generated negative samples. Specifically, we start with a prompt template to extract textual prior knowledge from the ChatGPT-3.5 model, enhancing the feature space for underrepresented classes. Then using this prior knowledge, a conditional diffusion model generates semantic-rich image samples for tail classes. Moreover, the proposed ChatDiff leverages a CLIP-based discriminator to screen and remove generated negative samples. This process avoids neural network learning the invalid or erroneous features, and further, improves long-tailed classification performance. Comprehensive experiments conducted on long-tailed benchmarks such as CIFAR10-LT, CIFAR100-LT, ImageNet-LT, and iNaturalist 2018, validate the effectiveness of our ChatDiff method.

Prior Knowledge, Acceptance, Adaptation, and Challenges Following Stoma Formation among Colorectal Cancer Patients in Northern Peninsular of Malaysia: A Qualitative Study

INTRODUCTION: Stoma formation affects an individual in various ways, including physical, emotional, social, and cognitive functions. Diverse studies report ways of an individual lives with new stoma formation. However, the comprehensive understanding of the entire process by the patient, which includes knowledge before the surgical procedure, as well as the subsequent acceptance, adaptation, and challenges to living with a stoma is lacking. MATERIALS AND METHODS: In-depth interview session were conducted with 12 colorectal cancer patients who have undergone surgical procedures for intestinal stoma formation. The patterns and themes within the data were identified by thematic analysis, involving data familiarisation and coding followed by themes’ generation and refinement of the themes. RESULTS: Four themes and 9 subthemes were identified, which revealed the sufficiency of stoma-related information and understanding prior to surgery as well as positive acceptance of self and family members reflected through their reactions and support. Nonetheless, the challenges were anticipated which highlights the complications of the stoma itself, obstacles surrounding social life, and financial burdens. CONCLUSION: This study provided valuable insights into the experiences of individuals living with a stoma following colorectal cancer surgery. The themes and subthemes highlight the need to address social stigma as well as financial issues to alleviate the burden of stoma-related expenses. Increasing public awareness and improving financial assistance could be a way to enhance the overall quality of life for patients living with stoma

Bayesian Direction-of-Arrival Estimation Using Atomic Norm Minimization With Prior Knowledge

Bayesian Direction-of-Arrival Estimation Using Atomic Norm Minimization With Prior Knowledge

FuBay: An Integrated Fusion Framework for Hyperspectral Super-Resolution Based on Bayesian Tensor Ring.

Fusion with corresponding finer-resolution images has been a promising way to enhance hyperspectral images (HSIs) spatially. Recently, low-rank tensor-based methods have shown advantages compared with other kind of ones. However, these current methods either relent to blind manual selection of latent tensor rank, whereas the prior knowledge about tensor rank is surprisingly limited, or resort to regularization to make the role of low rankness without exploration on the underlying low-dimensional factors, both of which are leaving the computational burden of parameter tuning. To address that, a novel Bayesian sparse learning-based tensor ring (TR) fusion model is proposed, named as FuBay. Through specifying hierarchical sprasity-inducing prior distribution, the proposed method becomes the first fully Bayesian probabilistic tensor framework for hyperspectral fusion. With the relationship between component sparseness and the corresponding hyperprior parameter being well studied, a component pruning part is established to asymptotically approaching true latent rank. Furthermore, a variational inference (VI)-based algorithm is derived to learn the posterior of TR factors, circumventing nonconvex optimization that bothers the most tensor decomposition-based fusion methods. As a Bayesian learning methods, our model is characterized to be parameter tuning-free. Finally, extensive experiments demonstrate its superior performance when compared with state-of-the-art methods.

Biologically weighted LASSO: enhancing functional interpretability in gene expression data analysis.

Feature selection approaches are widely used in gene expression data analysis to identify the most relevant features and boost performance in regression and classification tasks. However, such algorithms solely consider each feature's quantitative contribution to the task, possibly limiting the biological interpretability of the results. Feature-related prior knowledge, such as functional annotations and pathways information, can be incorporated into feature selection algorithms to potentially improve model performance and interpretability. We propose an embedded integrative approach to feature selection that combines weighted LASSO feature selection and prior biological knowledge in a single step, by means of a novel score of biological relevance that summarizes information extracted from popular biological knowledge bases. Findings from the performed experiments indicate that our proposed approach is able to identify the most predictive genes while simultaneously enhancing the biological interpretability of the results compared to the standard LASSO regularized model. Code is available at https://github.com/DEIB-GECO/GIS-weigthed_LASSO.

Learning Constrained Dynamic Correlations in Spatiotemporal Graphs for Motion Prediction.

Human motion prediction is challenging due to the complex spatiotemporal feature modeling. Among all methods, graph convolution networks (GCNs) are extensively utilized because of their superiority in explicit connection modeling. Within a GCN, the graph correlation adjacency matrix drives feature aggregation, and thus, is the key to extracting predictive motion features. State-of-the-art methods decompose the spatiotemporal correlation into spatial correlations for each frame and temporal correlations for each joint. Directly parameterizing these correlations introduces redundant parameters to represent common relations shared by all frames and all joints. Besides, the spatiotemporal graph adjacency matrix is the same for different motion samples, and thus, cannot reflect samplewise correspondence variances. To overcome these two bottlenecks, we propose dynamic spatiotemporal decompose GC (DSTD-GC), which only takes 28.6% parameters of the state-of-the-art GC. The key of DSTD-GC is constrained dynamic correlation modeling, which explicitly parameterizes the common static constraints as a spatial/temporal vanilla adjacency matrix shared by all frames/joints and dynamically extracts correspondence variances for each frame/joint with an adjustment modeling function. For each sample, the common constrained adjacency matrices are fixed to represent generic motion patterns, while the extracted variances complete the matrices with specific pattern adjustments. Meanwhile, we mathematically reformulate GCs on spatiotemporal graphs into a unified form and find that DSTD-GC relaxes certain constraints of other GC, which contributes to a better representation capability. Moreover, by combining DSTD-GC with prior knowledge like body connection and temporal context, we propose a powerful spatiotemporal GCN called DSTD-GCN. On the Human3.6M, Carnegie Mellon University (CMU) Mocap, and 3D Poses in the Wild (3DPW) datasets, DSTD-GCN outperforms state-of-the-art methods by 3.9%-8.7% in prediction accuracy with 55.0%-96.9% fewer parameters. Codes are available at https://github.com/Jaakk0F/DSTD-GCN.

Top-down attention shifts behavioral and neural event boundaries in narratives with overlapping event scripts

Understanding and remembering the complex experiences of everyday life relies critically on prior schematic knowledge about how events in our world unfold over time. How does the brain construct event representations from a library of schematic scripts, and how does activating a specific script impact the way that events are segmented in time? We developed a novel set of 16 audio narratives, each of which combines one of four location-relevant event scripts (restaurant, airport, grocery store, and lecture hall) with one of four socially relevant event scripts (breakup, proposal, business deal, and meet cute), and presented them to participants in an fMRI study and a separate online study. Responses in the angular gyrus, parahippocampal gyrus, and subregions of the medial prefrontal cortex (mPFC) were driven by scripts related to both location and social information, showing that these regions can track schematic sequences from multiple domains. For some stories, participants were primed to attend to one of the two scripts by training them to listen for and remember specific script-relevant episodic details. Activating a location-related event script shifted the timing of subjective event boundaries to align with script-relevant changes in the narratives, and this behavioral shift was mirrored in the timing of neural responses, with mPFC event boundaries (identified using a hidden Markov model) aligning to location-relevant rather than socially relevant boundaries when participants were location primed. Our findings demonstrate that neural event dynamics are actively modulated by top-down goals and provide new insight into how narrative event representations are constructed through the activation of temporally structured prior knowledge.

Neural network algorithm with transfer learning and dropout for using a UAV to search the lost target in motion

When performing the rescue mission, how to find a lost target in motion as soon as possible is a very valuable and interesting research topic. To find the optimal flight path of the UAV used to search the lost target in motion under different prior knowledge and interference intensity, this paper proposes a novel optimization technique called neural network algorithm with transfer learning and dropout (TLDNNA), which is a variant of neural network algorithm (NNA) inspired by artificial neural networks. To improve the global search ability and convergence performance of NNA, the multiple transfer mechanism, the generalized mean transfer position, and the random dropout operator driven by the principles of transfer learning and dropout are introduced to TLDNNA. To investigate the performance of TLDNNA, TLDNNA and 12 powerful population-based optimization algorithms are applied to the problem of using a UAV to search the lost target in motion in the designed 9 scenarios. Experimental results show that TLDNNA is remarkably better than the other 12 algorithms in most of the scenarios in terms of solution quality, stability, and convergence performance. In addition, TLDNNA can improve the overall search performance by 18.14% and overall computational efficiency by 3.41 times compared to NNA. The source code of TLDNNA can be obtained by https://github.com/jsuzyy/TLDNNA.