Average Improvement Research Articles

Using experience classification for training non-Markovian tasks

Unlike standard Reinforcement Learning (RL) model, many real-world tasks are non-Markovian, which requires long-term memory and dependency. Hence solving a non-Markovian task is more difficult and challenging than solving a Markovian one. In this paper, we propose a novel RL approach for training non-Markovian tasks expressed in temporal logic LTLf (Linear Temporal Logic over Finite Traces). To this end, an encoding of linear complexity from LTLf into MDPs (Markov Decision Processes) is introduced in order to take advantage of advanced RL algorithms. We further propose an experience classification method based on the automaton structure (theoretically equivalent to LTLf specifications). An automatic reward shaping technique and a prioritized experience replay mechanism are developed to cooperate with the classification method to improve the performance of RL algorithms. We provide empirical evaluations on two widely used benchmark problems, Waterworld and Cartpole, both augmented with complex non-Markovian tasks. The evaluations are conducted with respect to the metrics of training speed, policy quality, convergence rates, computational efficiency and scalability etc. The experimental results show that our approach achieves superior performance over other relevant studies, specially with an average improvement of 133% in convergence rate and a reduction of 11% in training time.

Deep Brain Stimulation of the Subthalamic Nucleus vs. Globus Pallidus Internus in the Treatment of Parkinson’s Disease: A Systematic Review

Abstract Background Parkinson’s disease (PD) is a very common debilitating disease especially in the elderly population. Bilateral stimulation of the subthalamic nucleus or the globus pallidus internus provides significant motor benefits for patients with advanced Parkinson’s disease, while reducing dyskinesia and motor fluctuation. High frequency stimulation of the STN and the GPi is relatively safe. The effects are stable even after 1 year of stimulation. Aim of the Work The primary goal of the current review is to compare the efficacy of using the Subthalamic Nucleus (STN) vs. the Globus Pallidus Internus (GPi) as the primary target for Deep Brain Stimulation (DBS) in the treatment of Parkinson disease (PD); in terms of improved motor function measured using the Unified Parkinson's Disease Rating Scale (UPDRS), psychiatric symptoms, rate of reduction of medication, and complications. This will be done through a systematic review of literature addressing this research question. Patients and Methods In this Review, the authors are aiming at the analysis of the collective outcomes from studies comparing bilateral deep brain stimulation of the Subthalamic Nucleus and the Globus Pallidus Internus in the management of idiopathic Parkinson’s disease. The following electronic databases were searched for included articles: PubMed, Google Scholar, EMBASE, Science Direct, Cochrane database of systematic reviews, using the keywords “Parkinson’s disease – treatment – management – deep brain stimulation – subthalamic – globus pallidus internus” in different combinations. Also, manual search were conducted for reference lists of included articles. Limiting the search in English. The information sources previously mentioned were searched from 2000 to 2021. Results A total of 594 patients underwent STN-DBS, with a mean change in total UPDRSIII score of 3.26 on medication and 25.04 off medication. A total of 446 patients underwent GPi-DBS, with a mean change in total UPDRSIII score of 4.75 on medication and 19.31 off medication. There was a higher change for GPi on medication that was statistically significant with a p value (˂0.001), while off medication STN caused a higher change that was also statistically significant with a p value (˂0.001). We observed a statistically significant higher average improvement in ADLs after STN-DBS than GPi, however, non- motor function improvements were greater after GPi DBS than STN DBS. Conclusion Based on our systematic review and other available data in the literature, the STN may be the preferred target for DBS in consideration of motor function improvement, medication reduction, and economic efficiency, although GPI stimulation has some advantages in terms of mood and cognition. Treatment decisions should be individualized and based on patients’ specific symptoms and expectations. Additional well-designed multicenter clinical trials that can minimize bias are needed, especially for evaluation of the long-term outcomes of the two stimulation targets.

Change in Mindfulness Profiles After Mindfulness-Based Cognitive Therapy for Major Depressive Disorder

ObjectivesMindfulness-Based Cognitive Therapy (MBCT) reduces recurrence and current depressive symptoms in patients with major depressive disorder (MDD). To understand how and for whom MBCT works, a person-centered approach focusing on mindfulness profiles can be useful. Four mindfulness profiles, each associated differently with mental health outcomes, have previously been identified. So far, no studies have examined whether profiles change after MBCT and whether these changes are related to treatment outcome.MethodLatent transition analysis (LTA) was performed on pre- and post-MBCT subscale scores of the Five Facet Mindfulness Questionnaire (FFMQ) in patients with current or remitted MDD (n=500). LTA allowed the assessment of individual changes in mindfulness profile after MBCT and the relation between profile change and corresponding changes in measures of mental health, including depressive symptoms, overall functional impairment, worry, and self-compassion.ResultsLTA re-established the four profiles previously identified cross-sectionally: “Very low mindfulness” (VLM), “Non-judgmentally aware” (NJA), “Judgmentally observing” (JO), and “High mindfulness” (HM). For 71 out of 168 patients with VLM profiles changed to NJA and for another 30 to HM. For 49 out of the 129 patients with NJA and for 37 out of 141 patients with JO profiles changed to HM. All 61 patients starting with HM kept HM. In general, change was related to greater than average improvement in mental health, while no change in profile was related to less than average improvement in mental health (except for HM).ConclusionsOur findings indicate that changes in mindfulness profile after MBCT was differentially related to clinical change. These results from a person-centered approach offer new avenues to further elucidate the working mechanism of MBCT and improve its outcome.PreregistrationThis study was not preregistered.

Precise point positioning (PPP) based on the machine learning-based ionospheric tomography

Undifferenced and Uncombined Precise Point Positioning (UPPP) currently stands as a prominent research area, where the integration of high-precision ionospheric products holds the potential to substantially enhance accuracy and convergence performance in UPPP. Presently, the majority of external ionospheric constraints for UPPP rely on Global Ionospheric Maps (GIM). However, the accuracy and resolution of GIM fall short, imposing significant limitations on the positioning performance of UPPP. Consequently, this paper introduces the application of Computerized Ionospheric Tomography based on Machine Learning (CIT-ML) to enhance UPPP performance (PPP-CIT-ML). In this approach, we convert the three-dimensional electron density of CIT-ML into vertical total electron content (VTEC), and then it is compiled into ionospheric grid files essential for UPPP. Simultaneously, the traditional ionospheric tomography methods based on the improved algebraic reconstruction technique (IART) is also employed for UPPP (PPP-CIT-IART), alongside the ionospheric grid files broadcasted by the Center for Orbit Determination in Europe (CODE) for UPPP (PPP-CODE), where PPP-CIT-IART and PPP-CODE are used as the reference methods to test the performance of PPP-CIT-ML. In Static and Kinematic UPPP, compared to PPP-CODE, PPP-CIT-IART demonstrated average improvements in positioning accuracy and convergence performance by over 10% and 7%, respectively. PPP-CIT-ML showed average improvements in positioning accuracy and convergence performance by over 26% and 28%, respectively. The extrapolated ionospheric electron density (IED) applied to UPPP (ECIT-ML-PPP) and compared with PPP-CODE displayed average improvements in positioning accuracy by over 21% and convergence performance by over 26%. Compared with PPP-CIT-IART, ECIT-ML-PPP displayed average improvements in positioning accuracy and convergence performance by over 6% and 21%, respectively. These findings highlight that ionospheric error correction information obtained through ionospheric tomography significantly enhances the positioning accuracy and convergence performance of UPPP. Moreover, the performance enhancement achieved through machine learning-based ionospheric tomography is more pronounced. This study provides preliminary validation for the feasibility of applying machine learning-based ionospheric tomography results to navigation positioning.

The Street Medicine Auto-Refraction Technology (SMART) Study: A Model for Screening and Treating Refractive Error in Persons Experiencing Homelessness

PurposeThe SMART study is an efficient and cost-effective treatment of refractive-error (RE) and identification of potential non-RE pathology in people experiencing homelessness (PEH). DesignOutreach evaluation for diagnosis and treatment. MethodsThe internal medicine physicians/staff team used a protocol designed by ophthalmology to screen and treat PEH. First, the uncorrected visual acuity (ucVA) was measured, which was defined as a patient's visual acuity without the use of any glasses. Those with 2-line improvement on pinhole testing (phVA) underwent handheld portable auto-refraction (AR). During a second encounter, glasses purchased from the AR measurements were distributed, and vision was re-measured with the new glasses or current corrected visual acuity (ccVA). Patients with a final VA ≥0.3 logMAR (equivalent to Early Treatment Diabetic Retinopathy Study (ETDRS) 20/40) were referred to an ophthalmology clinic for further evaluation. ResultsThis was a prospective study of 117 patients (226 eyes). ucVA was 0.64 logMAR (equivalent to ETDRS 20/87) for both the right (OD) and the left eye (OS). 90/98 (91.8%) patients successfully underwent AR. 80/88 (90.1%) patients have received their glasses. For the 61 patients that completed their second visit, the average improvement in VA (ccVA – ucVA) was 5 lines. 23/117 (19.7%) were referred to ophthalmology for further evaluation. Post-study Likert surveys (1=strongly disagree to 5=strongly agree) showed patient satisfaction with the testing (4.7), the AR (4.6), seeing better (4.8), and doing activities with glasses (4.7). Additionally, most patients agreed they would not have obtained glasses without the study (4.7). ConclusionsThis is a new model to screen RE and provide affordable glasses to PEH in a non-ambulatory setting. Patients with potential non-RE can be sent for further evaluation.

Data-driven topology optimization using a multitask conditional variational autoencoder with persistent homology

Topology optimization is crucial for the mechanical design of vehicles and aircraft, allowing changes in the shape of structures and the placement of features. Recent advances have integrated deep generative models, particularly convolutional neural networks, to streamline this process.to streamline this process. However, these models struggle to preserve subtle structural features. To overcome these limitations, this study introduced a generative model adept at identifying the topological features inherent in real shapes, such as connectivity and holes, to enhance the effectiveness of topology optimization. A conditional variational autoencoder (CVAE) was employed to predict both the shape and compliance simultaneously. This model, CVAE with persistent homology, generates optimal material distributions by considering topological properties. The learning process introduced a term that minimizes the difference in topological features between true and reconstructed shapes. The proposed model can generate optimal material distributions by considering topological properties, eliminating the need for iterative calculations. This approach was validated using two numerical examples. The accuracy of the generated material distributions was compared with conventional methods using the mean-squared error. An average improvement in accuracy of approximately 36.85% was observed across the two results. This confirms that shapes considering compliance and connectivity can be accurately predicted.

Multiobjective optimization of porous medium for efficient transpiration cooling of hypersonic vehicles using genetic algorithm

Transpiration cooling has been proven an effective method for reducing heat flux on the surfaces of high-speed vehicles. This study investigates the effects of porous medium properties, employing a black-box optimization method to determine the optimal length, thickness, and porosity for a porous medium in a transpiration cooling system on a flat plate under hypersonic laminar flow. The objectives include optimizing thermal effectiveness, coolant consumption, and the weight and cost of the porous material. A multiobjective genetic optimization algorithm is directly integrated with a Computational Fluid Dynamics (CFD) solver, and 1562 CFD simulations were conducted to identify the optimal configuration. The results demonstrate that the length and porosity of the porous medium more significantly impact thermal effectiveness than the thickness. Furthermore, the optimization identified a configuration for the porous medium that, when compared to the original case, shows reductions in length, thickness, and porosity of 3.5%, 11%, and 29%, respectively. Additionally, there were average improvements in thermal effectiveness and coolant consumption of 4.56% and 3.9%, respectively, while the weight and cost of the porous material increased by 3.73% and 3.65%, respectively.

Groundwater well rehabilitation using hydropuls method: a case study in wonosobo groundwater well

Abstract Hydropuls is one of the groundwater well rehabilitation methods that involves pulsing highly compressed nitrogen gas through screen pipes at an adjustable interval. Nitrogen gas was chosen because of its colorless, odorless, and inert properties. The aquifer lithology of this groundwater well is sand, gravel and breccias (member of Sundoro Volcanics; Qsu) with silty clay as the impermeable layer that makes the aquifer type is confined aquifer. The following methods were used in this well rehabilitation: step tests before, borehole camera before, hydropuls, airlifting, borehole camera after and step test after well rehabilitation. Thin section analysis shows that sediment mostly consists of lithic fragments and plagioclase. The results of the polished section analysis reveal that most of the metallic minerals are magnetite and hematite. The XRF results show that the sediment mostly consists of SiO2, Al2O3 and Fe2O3. According to the XRD results, the sediment is mostly albite. The interpretation of sediment analysis shows that the clogging is caused by iron precipitation. The result is an average increase in specific capacity of 1.43 l/s/m (a 168% increase over the initial conditions) and an average drawdown improvement of 2.56 m (a 114% improvement compared to pre-hydropuls conditions). Borehole camera imaging confirmed that the precipitation on the screen was removed and the screen is in good condition.

Improving the accuracy of genomic prediction in dairy cattle using the biologically annotated neural networks framework

BackgroundBiologically annotated neural networks (BANNs) are feedforward Bayesian neural network models that utilize partially connected architectures based on SNP-set annotations. As an interpretable neural network, BANNs model SNP and SNP-set effects in their input and hidden layers, respectively. Furthermore, the weights and connections of the network are regarded as random variables with prior distributions reflecting the manifestation of genetic effects at various genomic scales. However, its application in genomic prediction has yet to be explored.ResultsThis study extended the BANNs framework to the area of genomic selection and explored the optimal SNP-set partitioning strategies by using dairy cattle datasets. The SNP-sets were partitioned based on two strategies–gene annotations and 100 kb windows, denoted as BANN_gene and BANN_100kb, respectively. The BANNs model was compared with GBLUP, random forest (RF), BayesB and BayesCπ through five replicates of five-fold cross-validation using genotypic and phenotypic data on milk production traits, type traits, and one health trait of 6,558, 6,210 and 5,962 Chinese Holsteins, respectively. Results showed that the BANNs framework achieves higher genomic prediction accuracy compared to GBLUP, RF and Bayesian methods. Specifically, the BANN_100kb demonstrated superior accuracy and the BANN_gene exhibited generally suboptimal accuracy compared to GBLUP, RF, BayesB and BayesCπ across all traits. The average accuracy improvements of BANN_100kb over GBLUP, RF, BayesB and BayesCπ were 4.86%, 3.95%, 3.84% and 1.92%, and the accuracy of BANN_gene was improved by 3.75%, 2.86%, 2.73% and 0.85% compared to GBLUP, RF, BayesB and BayesCπ, respectively across all seven traits. Meanwhile, both BANN_100kb and BANN_gene yielded lower overall mean square error values than GBLUP, RF and Bayesian methods.ConclusionOur findings demonstrated that the BANNs framework performed better than traditional genomic prediction methods in our tested scenarios, and might serve as a promising alternative approach for genomic prediction in dairy cattle.

Electrical performance and reliability analysis of vertical gallium nitride Schottky barrier diodes with dual-ion implanted edge termination

In this study, a gallium nitride (GaN) substrate and its 15 μm epitaxial layer were entirely grown using hydride vapor phase epitaxy (HVPE). To enhance the breakdown voltage (VBR) of vertical GaN-on-GaN Schottky Barrier Diodes (SBDs), a dual ion co-implantation of carbon and helium was used to create the edge termination. The resulting devices exhibited low turn-on voltage of 0.55 V, high Ion/Ioff of approximately 109, and low specific on-resistance of 1.93 mΩ·cm2. With the ion implantation edge termination, the devices achieved a maximum VBR of 1575 V, with an average improvement of 126%. These devices demonstrated a high figure of merit (FOM) of 1.28 GW/cm2 and showed excellent reliability during pulse stress testing.

Enhance carbon emission prediction using bidirectional long short-term memory model based on text-based and data-driven multimodal information fusion

Accurate projections of carbon emissions are essential for climate change responses and environmental policy development. However, the complexity of carbon emission projections has increased due to various factors like technology and policy, posing challenges for accuracy. Many studies rely on time-series data alone, overlooking the combined impact of multiple information sources on carbon emissions. To overcome these bottlenecks, a hybrid model is proposed, integrating multimodal data, convolutional neural network inceptionv2 (CNN-inceptionv2), text convolutional neural network (TextCNN) optimized by attention mechanism, and bi-directional long and short-term memory (BiLSTM). Firstly, the time series feature set and text feature set are extracted using the CNN-inceptionv2 and the TextCNN, respectively. Subsequently, the feature sets of the two modal information are fused and outputted into a multimodal feature set, which are ultimately predicted using BiLSTM. The validity of the study model was also verified by different methods such as ablation experiments, single model and combined model. The coefficient of determination (R2) values were 0.986 for the validation set and 0.933 for the prediction set. In comparison to other models, the present model showed an average improvement of 16.6% in R2 values, fully indicating its superior predictive performance. The high point and interval predictions of carbon emissions further verified the superior prediction accuracy of the present model. The present model can accurately predict carbon emissions, providing data references for carbon emission decision makers. Additionally, it further enriches the methodological system of carbon emission prediction and aids in strategic planning for future carbon emission and energy management.

Multipattern Mining Using Pattern-Level Contrastive Learning and Multipattern Activation Map.

Visual patterns are basic elements in images and represent the discernible regularity in the visual world. Thus, mining visual patterns is a fundamental task in computer vision. Most previous studies consider that only one visual pattern exists in a category, and then builds up a one-to-one mapping using category label. In reality, however, many categories include multiple patterns, which are many-to-one mappings. Without knowing the information of patterns, few existing pattern mining methods can discover and distinguish varied patterns in a category. To tackle this problem, we propose a novel framework, PaclMap, which learns medium-grained features to represent patterns. It includes an unsupervised pattern-level contrastive learning and a multipattern activation map. Their joint optimization encourages the network to mine both discriminative and frequent patterns in a category. Extensive experiments conducted on four benchmark datasets (Place-20, imagenet large scale visual recognition challenge (ILSVRC)-20, visual object classes (VOC), and Travel) demonstrate that PaclMap outperforms six state-of-the-art methods with average improvements of 2.9% on accuracy and 12.3% on frequency, respectively.

PMTT: Parallel multi-scale temporal convolution network and transformer for predicting the time to aging failure of software systems

Software aging is one of the significant factors affecting the reliability and availability of long-running software systems, such as Android, Cloud systems, etc. The time to aging failure (TTAF) prediction for software systems plays a crucial role in proactive rejuvenation scheduling through machine learning or statistical analysis techniques, due to its ability to determine when to perform rejuvenation to mitigate the aging effects. However, software aging characterization is relatively complicated, and only fitting the variations for a single aging indicator cannot grasp the comprehensive degradation process across different case systems; moreover, since software systems often exhibit long and short-term inherent degradation characteristics, existing prediction models possess a poor ability for modeling both global and local information simultaneously. To tackle the above problems, a novel TTAF prediction framework based on the parallel multi-scale temporal convolution network and transformer (named PMTT) is proposed, by mapping various system running indicators reflecting the software aging to TTAF. PMTT possesses the following distinctive characteristics. First, a local feature extraction module that contains multiple channel TCNs with different scales is developed to extract inherent local information from the raw input. Second, in a parallel manner, a global feature extraction module integrating transformer blocks is built to extract global information representation synchronously using the self-attention mechanism. Afterward, high-level global–local features extracted from different channels are fused, and TTAF is estimated through two fully connected regression layers using the fused features. The proposed PMTT has been compared to seven competitors using run-to-failure data collected from Android and OpenStack systems. The experiments have demonstrated the superiority of PMTT, showing an average improvement of 11.2%, 9.0%, and 9.3% in performance across three evaluation metrics compared with the optimal baseline model.

Combining Neuromodulation Strategies in Spinal Cord Injury Gait Rehabilitation: A Proof of Concept, Randomized, Crossover Trial

ObjectivesTo evaluate if acute intermittent hypoxia (AIH) coupled with transcutaneous spinal cord stimulation (tSCS) enhances task-specific training and leads to superior and more sustained gait improvements as compared with each of these strategies used in isolation in persons with chronic, incomplete spinal cord injury. DesignProof of concept, randomized crossover trial. SettingOutpatient, rehabilitation hospital. InterventionsTen participants completed 3 intervention arms: (1) AIH, tSCS, and gait training (AIH + tSCS); (2) tSCS plus gait training (SHAM AIH + tSCS); and (3) gait training alone (SHAM + SHAM). Each arm consisted of 5 consecutive days of intervention with a minimum of a 4-week washout between arms. The order of arms was randomized. The study took place from December 3, 2020, to January 4, 2023. Main Outcome Measures10-meter walk test at self-selected velocity (SSV) and fast velocity, 6-minute walk test, timed Up and Go (TUG) and secondary outcome measures included isometric ankle plantarflexion and dorsiflexion torque ResultsTUG improvements were 3.44 seconds (95% CI: 1.24-5.65) significantly greater in the AIH + tSCS arm than the SHAM AIH + tSCS arm at post-intervention (POST), and 3.31 seconds (95% CI: 1.03-5.58) greater than the SHAM + SHAM arm at 1-week follow up (1WK). SSV was 0.08 m/s (95% CI: 0.02-0.14) significantly greater following the AIH + tSCS arm than the SHAM AIH + tSCS at POST. Although not significant, the AIH + tSCS arm also demonstrated the greatest average improvements compared with the other 2 arms at POST and 1WK for the 6-minute walk test, fast velocity, and ankle plantarflexion torque. ConclusionsThis pilot study is the first to demonstrate that combining these 3 neuromodulation strategies leads to superior improvements in the TUG and SSV for individuals with chronic incomplete spinal cord injury and warrants further investigation.

AEMNet: Unsupervised Video Anomaly Detection Method Based on Attention-Enhanced Memory Networks

Video anomaly detection has always been a challenging task in computer vision due to data imbalance and susceptibility to scene variations such as lighting and occlusions. In response to this challenge, this paper proposes an unsupervised video anomaly detection method based on an attention-enhanced memory network. The method utilizes a dual-stream network structure of autoencoders, enhancing the model’s learning ability for important features in appearance and motion by introducing coordinate attention mechanisms and variance attention mechanisms, emphasizing significant characteristics of static objects and rapidly moving regions. By adding memory modules to both the appearance and motion branches, the network structure’s memory information is reinforced, enabling it to capture long-term spatiotemporal dependencies in videos and thereby improving the accuracy of anomaly detection. Furthermore, by optimizing the network structure’s activation functions to handle negative inputs, it enhances its nonlinear modeling capabilities, enabling better adaptation to complex environments, including variations in lighting and occlusions, further improving the effectiveness of anomaly detection. The paper conducts comparative experiments and ablation studies using three public available datasets and various models. The results demonstrate that compared to baseline models, the AUC performance is improved by 3.9%, 4.7%, and 1.7% on UCSD Ped2, CHUK Avenue, and ShanghaiTech datasets, respectively. When compared with the other models, the average AUC performance is improved by 4.3%, 5.4%, and 6.2%, with an average improvement of 8.75% in the ERR metric, validating the effectiveness and adaptability of the proposed method. The code can be obtained at the following URL: https://github.com/AcademicWhite/AEMNet .

Feature selection for multi-labeled data based on label enhancement technique and mutual information

In multi-label data, the importance of each label within the logical label vector varies for each sample, and there exist inherent correlations among the labels. However, the logical label vector fails to capture these nuances. Consequently, relying solely on this vector for feature selection in multi-label data results in underutilization of supervisory information. To address this issue, this paper introduces a novel label enhancement algorithm. This algorithm leverages neighborhood information derived from features to transform the logical label vector into a label distribution that effectively reflects label differences and correlations. Subsequently, we propose a feature selection algorithm tailored for multi-label data, which incorporates both the transformed label distribution and mutual information. This algorithm not only accounts for the mutual information between features and label distributions but also captures the mutual information among features themselves. Finally, we evaluate our proposed feature selection algorithm against five state-of-the-art multi-label feature selection algorithms on ten publicly available datasets. The experimental results reveal that our algorithm outperforms its competitors in six distinct evaluation metrics, achieving an average performance improvement of approximately 9%. This substantial enhancement underscores the efficacy of our algorithm in handling complex multi-label data.

RiboDiffusion: tertiary structure-based RNA inverse folding with generative diffusion models.

RNA design shows growing applications in synthetic biology and therapeutics, driven by the crucial role of RNA in various biological processes. A fundamental challenge is to find functional RNA sequences that satisfy given structural constraints, known as the inverse folding problem. Computational approaches have emerged to address this problem based on secondary structures. However, designing RNA sequences directly from 3D structures is still challenging, due to the scarcity of data, the nonunique structure-sequence mapping, and the flexibility of RNA conformation. In this study, we propose RiboDiffusion, a generative diffusion model for RNA inverse folding that can learn the conditional distribution of RNA sequences given 3D backbone structures. Our model consists of a graph neural network-based structure module and a Transformer-based sequence module, which iteratively transforms random sequences into desired sequences. By tuning the sampling weight, our model allows for a trade-off between sequence recovery and diversity to explore more candidates. We split test sets based on RNA clustering with different cut-offs for sequence or structure similarity. Our model outperforms baselines in sequence recovery, with an average relative improvement of 11% for sequence similarity splits and 16% for structure similarity splits. Moreover, RiboDiffusion performs consistently well across various RNA length categories and RNA types. We also apply in silico folding to validate whether the generated sequences can fold into the given 3D RNA backbones. Our method could be a powerful tool for RNA design that explores the vast sequence space and finds novel solutions to 3D structural constraints. The source code is available at https://github.com/ml4bio/RiboDiffusion.

Detection of Straw Coverage under Conservation Tillage Based on an Improved Mask Regional Convolutional Neural Network (Mask R-CNN)

Conservation tillage, a crucial method for protecting soil fertility, depends heavily on maintaining adequate straw coverage. The current method of straw coverage detection relies primarily on manual measurement, which is both time-consuming and laborious. This paper introduces a novel straw coverage detection approach based on an improved mask regional convolutional neural network (Mask R-CNN) algorithm. Several images of wheat straw-covered fields were taken, and the dataset was augmented using techniques like image inversion, contrast enhancement, Gaussian noise addition, and translation after cropping the original images. These fields use a crop rotation cycle of wheat and corn. Subsequently, the straw images were annotated using the Labelme annotation tool to obtain the available straw instance segmentation dataset. The Mask R-CNN algorithm was improved by refining the mask generation network structure through a multi-feature fusion strategy, which interweaves features from both the encoder and the mask generation network, enhancing the model’s ability to capture detailed and shape information of the straw. Lastly, using the mask information output by the improved Mask R-CNN algorithm, the straw coverage was calculated by counting the proportion of pixels within each segmented region. In the results, compared to the original Mask R-CNN algorithm, our improved Mask R-CNN algorithm achieved an average improvement of 7.8% in segmentation accuracy, indicating that the improved Mask R-CNN algorithm offers superior segmentation performance. Thus, the new algorithm can achieve straw coverage detection with higher accuracy and can provide a reference for other agricultural applications.

Assessment of green biocomposites based on polyvinyl alcohol and grafted biopolymer tamarind kernel powder with polyacrylic acid

This investigation focuses on the development and assessment of biopolymer-based green biocomposite materials that contain tamarind seed kernel powder (TKP) either in its raw or polyacrylic acid (PAA) grafted form as the reinforcing component, combined with polyvinyl alcohol (PVA) as the matrix material. The goal is to develop biopolymer TKP based green biocomposites with a biodegradable PVA matrix for food grade packaging and biomedical applications. Both raw TKP reinforced PVA and PAA grafted TKP reinforced PVA, two forms of sustainable green biocomposites, were developed. Compression molding was used to combine various raw or PAA grafted TKP loadings with PVA to produce various biocomposites. The loadings were 0 %, 0.25 %, 0.5 %, 1 %, 5 %, and 10 % in weight. To confirm the PAA grafted TKP, attenuated total reflection-fourier transform infrared (ATR-FTIR) spectrum analysis was employed. The mechanical properties of the biocomposites were examined, and it was discovered that the grafted biocomposites exhibited 4 % higher tensile strength and elongation at break (%) on average than the raw TKP reinforced PVA biocomposites. The Charpy impact strength of the biocomposites exhibited an average 28 % improvement in impact resistance with modified TKP and an 8 % increase with raw TKP when compared to pure PVA sheets. Additionally, the PAA-grafted TKP reinforced PVA biocomposites outperformed the raw TKP reinforced PVA biocomposites in water absorption tests, exhibiting an average improvement in moisture resistance of nearly 13 %. Scanning electron microscopy (SEM) tests revealed that the PAA-grafted biocomposites had a more uniform dispersion of fillers inside the PVA matrix as compared to the raw TKP/PVA biocomposites. These findings enable in the development of high-performance, environmentally friendly and green biocomposite materials for a variety of uses.

Addressing imbalance in graph datasets: Introducing GATE-GNN with graph ensemble weight attention and transfer learning for enhanced node classification

Significant challenges arise when Graph Neural Networks (GNNs) try to deal with uneven data. Specifically in signed and weighted graph structures. This makes classification tasks less effective. Within the GNN context, researchers have found traditional solutions like resampling, reweighting, and synthetic sample generation to be inadequate. GATE-GNN, a novel architecture designed specifically for imbalanced datasets, overcomes these limitations. GATE-GNN integrates an ensemble of network modules that harness the spatial features of graph networks and effectively utilise embedding information from earlier layers. This unique approach not only bolsters generalisation by reducing volatility. It also refines the optimisation algorithm, resulting in more accurate and stable classification outcomes. We rigorously tested the effectiveness of GATE-GNN on four widely recognised datasets: Cora, NELL, Citeseer, and PubMed. We performed a comparative analysis against established methods such as Graph Convolutional Networks (GCN), Graph Sample and Aggregate (GraphSAGE), Propagation Multilayer Perceptron PMLP), Imbalanced Node Sampling GNN (INS-GNN), GNN-Curriculum Learning (GNN-CL) and Graph Attention Networks (GAT). Empirical results demonstrate that GATE-GNN significantly outperforms these existing models, achieving an average improvement in classification accuracy of approximately 5%–10% over the previous best results. Additionally, GATE-GNN presents a marked reduction in training time. This underscores its efficiency and suitability for practical applications in imbalanced graph data scenarios. Implementation of the proposed GATE-GNN can be accessed here https://github.com/afofanah/GATE-GNN.