Extraction Framework Research Articles

HSF-IBI: A Universal Framework for Extracting Inter-Beat Interval from Heterogeneous Unobtrusive Sensors

Heartbeat inter-beat interval (IBI) extraction is a crucial technology for unobtrusive vital sign monitoring, yet its precision and robustness remain challenging. A promising approach is fusing heartbeat signals from different types of unobtrusive sensors. This paper introduces HSF-IBI, a novel and universal framework for unobtrusive IBI extraction using heterogeneous sensor fusion. Specifically, harmonic summation (HarSum) is employed for calculating the average heart rate, which in turn guides the selection of the optimal band selection (OBS), the basic sequential algorithmic scheme (BSAS)-based template group extraction, and the template matching (TM) procedure. The optimal IBIs are determined by evaluating the signal quality index (SQI) for each heartbeat. The algorithm is morphology-independent and can be adapted to different sensors. The proposed algorithm framework is evaluated on a self-collected dataset including 19 healthy participants and an open-source dataset including 34 healthy participants, both containing heterogeneous sensors. The experimental results demonstrate that (1) the proposed framework successfully integrates data from heterogeneous sensors, leading to detection rate enhancements of 6.25 % and 5.21 % on two datasets, and (2) the proposed framework achieves superior accuracy over existing IBI extraction methods, with mean absolute errors (MAEs) of 5.25 ms and 4.56 ms on two datasets.

FastQAFPN-YOLOv8s-Based Method for Rapid and Lightweight Detection of Walnut Unseparated Material

Walnuts possess significant nutritional and economic value. Fast and accurate sorting of shells and kernels will enhance the efficiency of automated production. Therefore, we propose a FastQAFPN-YOLOv8s object detection network to achieve rapid and precise detection of unsorted materials. The method uses lightweight Pconv (Partial Convolution) operators to build the FasterNextBlock structure, which serves as the backbone feature extractor for the Fasternet feature extraction network. The ECIoU loss function, combining EIoU (Efficient-IoU) and CIoU (Complete-IoU), speeds up the adjustment of the prediction frame and the network regression. In the Neck section of the network, the QAFPN feature fusion extraction network is proposed to replace the PAN-FPN (Path Aggregation Network—Feature Pyramid Network) in YOLOv8s with a Rep-PAN structure based on the QARepNext reparameterization framework for feature fusion extraction to strike a balance between network performance and inference speed. To validate the method, we built a three-axis mobile sorting device and created a dataset of 3000 images of walnuts after shell removal for experiments. The results show that the improved network contains 6071008 parameters, a training time of 2.49 h, a model size of 12.3 MB, an mAP (Mean Average Precision) of 94.5%, and a frame rate of 52.1 FPS. Compared with the original model, the number of parameters decreased by 45.5%, with training time reduced by 32.7%, the model size shrunk by 45.3%, and frame rate improved by 40.8%. However, some accuracy is sacrificed due to the lightweight design, resulting in a 1.2% decrease in mAP. The network reduces the model size by 59.7 MB and 23.9 MB compared to YOLOv7 and YOLOv6, respectively, and improves the frame rate by 15.67 fps and 22.55 fps, respectively. The average confidence and mAP show minimal changes compared to YOLOv7 and improved by 4.2% and 2.4% compared to YOLOv6, respectively. The FastQAFPN-YOLOv8s detection method effectively reduces model size while maintaining recognition accuracy.

A framework for flood inundation extraction based on microwave and optical remote sensing images

IntroductionEffective monitoring and evaluation of floodwaters are essential for disaster prevention and mitigation. The flood inundation range can be obtained by using traditional simulation methods, but these methods still have shortcomings. This work proposes an optimization method for traditional methods.MethodsThis study aims to introduce an effective solution for the rapid and accurate extraction of flood inundation areas, emphasizing the enhancement of extraction speed and dynamic monitoring throughout the flood event. The solution uses a normalized difference water index (NDWI), a refined threshold method, and a filtering process for microwave (radar) images. Sentinel-1 SAR (Synthetic Aperture Radar) and Sentinel-2 MSI (Multi-spectral Image) images served as the primary data sources. The Sentinel-2 images were preprocessed to extract pre-flood water bodies, while the Sentinel-1 SAR images were processed using the proposed filtering method to identify post-flood inundation areas.ResultsThe application and validation of this framework are demonstrated through the case of the 2020 flood event in Tongling, Anhui Province. The framework’s performance was validated through comparison with ground truth data, yielding high kappa accuracies of 98% for optical images and 89% for Synthetic Aperture Radar. The findings highlight the framework’s ability to capture high-accuracy changes in flood inundation areas and to characterize the dynamic process of flood inundation area changes.DiscussionThis study contributes to the field by enhancing the extraction speed and scope of water bodies from SAR images and improving the quality of microwave remote sensing data processing. It offers valuable insights for emergency rapid response and situational awareness in the context of extreme weather events and associated flood disasters.

Pyridine-based Ionic sp2 carbon-conjugated covalent organic frameworks for selective extraction of Pu(IV) from high-level liquid waste

Pyridine-based Ionic sp2 carbon-conjugated covalent organic frameworks for selective extraction of Pu(IV) from high-level liquid waste

A Bidirectional Extraction-Then-Evaluation Framework for Complex Relation Extraction

A Bidirectional Extraction-Then-Evaluation Framework for Complex Relation Extraction

Granular intents learning via mutual information maximization for knowledge-aware recommendation

Knowledge-aware recommender systems, which utilize knowledge graphs (KGs) to enrich item information, have been shown to improve the accuracy and explainability of recommendations. Besides, KGs are further explored to determine the intent of choosing items (i.e., the reason why users select items of interest). Conventional methods represent intents either as sets of relations in a KG or as KG entities. However, such approaches fail to fully leverage the combined information provided by both entities and relations. To address this issue, we propose a new KG-based user Intent Extraction Framework (KIEF) to capture user intents at a more fine-grained level for recommendation. Specifically, we propose a novel intent representation constructed with relation-aware entity representation, encouraging finer granularity for user intents. Furthermore, since a KG may contain noisy information that impairs the quality of user intent, it is compulsory to consider which factors in a KG are important to represent a user’s intent. Thus, we introduce global intent which are comprehensive features for the entire interactions of all users and local intent, which are empirical features of individual users from personal history. By maximizing mutual information between global and local intents, KIEF captures user preference for items. Through extensive experiments on four real-world benchmark datasets, we prove the superior performance of KIEF over the state-of-the-art and analyze interpretable explanations for understanding user intents.

GPU Accelerated MapReduce-Based Distributed Framework for Knowledge Extraction from Large Uncertain Data

GPU Accelerated MapReduce-Based Distributed Framework for Knowledge Extraction from Large Uncertain Data

Deciphering cell to cell spatial relationship for pathology images using SpatialQPFs

Understanding spatial dynamics within tissue microenvironments is crucial for deciphering cellular interactions and molecular signaling in living systems. These spatial characteristics govern cell distribution, extracellular matrix components, and signaling molecules, influencing local biochemical and biophysical conditions. Despite significant progress in analyzing digital pathology images, current methods for capturing spatial relationships are limited. They often rely on specific spatial features that only partially describe the complex spatial distributions of cells and are frequently tied to particular outcomes within predefined model frameworks. Furthermore, these methods are typically limited to field of view analysis, which restricts their capacity to capture spatial patterns across whole-slide images, thereby limiting their ability to fully address the complexities of tissue architecture. To address these limitations, we present SpatialQPFs (Spatial Quantitative Pathology Features), an R package designed to extract interpretable spatial features from cell imaging data using spatial statistical methodologies. Leveraging segmented cell information, our package offers a comprehensive toolkit for applying a range of spatial statistical methods within a stochastic process framework, including analyses of point process data, areal data, and geostatistical data. By decoupling feature extraction from specific outcome models, SpatialQPFs enables thorough large-scale spatial analyses applicable across diverse clinical and biological contexts. This approach enhances the depth and accuracy of spatial insights derived from tissue data, empowering researchers to conduct comprehensive spatial analyses efficiently and reproducibly. By providing a flexible and robust framework for spatial feature extraction, SpatialQPFs facilitates advanced spatial analyses, paving the way for new discoveries in tissue biology and pathology. SpatialQPFs code and documentation are publicly available at https://github.com/Genentech/SpatialQPFs.

Enhancing brain tumor MRI classification with an ensemble of deep learning models and transformer integration

Brain tumors are widely recognized as the primary cause of cancer-related mortality globally, necessitating precise detection to enhance patient survival rates. The early identification of brain tumor is presented with significant challenges in the healthcare domain, necessitating the implementation of precise and efficient diagnostic methodologies. The manual identification and analysis of extensive MRI data are presented as a challenging and laborious task, compounded by the importance of early tumor detection in reducing mortality rates. Prompt initiation of treatment hinges upon identifying the specific tumor type in patients, emphasizing the urgency for a dependable deep learning methodology for precise diagnosis. In this research, a hybrid model is presented which integrates the strengths of both transfer learning and the transformer encoder mechanism. After the performance evaluation of the efficacy of six pre-existing deep learning model, both individually and in combination, it was determined that an ensemble of three pretrained models achieved the highest accuracy. This ensemble, comprising DenseNet201, GoogleNet (InceptionV3), and InceptionResNetV2, is selected as the feature extraction framework for the transformer encoder network. The transformer encoder module integrates a Shifted Window-based Self-Attention mechanism, sequential Self-Attention, with a multilayer perceptron layer (MLP). These experiments were conducted on three publicly available research datasets for evaluation purposes. The Cheng dataset, BT-large-2c, and BT-large-4c dataset, each designed for various classification tasks with differences in sample number, planes, and contrast. The model gives consistent results on all three datasets and reaches an accuracy of 99.34%, 99.16%, and 98.62%, respectively, which are improved compared to other techniques.

Fine-Grained Meetup Events Extraction Through Context-Aware Event Argument Positioning and Recognition

Extracting meetup events from social network posts or webpage announcements is the core technology to build event search services on the Web. While event extraction in English achieves good performance in sentence-level evaluation [1], the quality of auto-labeled training data via distant supervision is not good enough for word-level event extraction due to long event titles [2]. Additionally, meetup event titles are more complex and diverse than trigger-word-based event extraction. Therefore, the performance of event title extraction is usually worse than that of traditional named entity recognition (NER). In this paper, we propose a context-aware meetup event extraction (CAMEE) framework that incorporates a sentence-level event argument positioning model to locate event fields (i.e., title, venue, dates, etc.) within a message and then perform word-level event title, venue, and date extraction. Experimental results show that adding sentence-level event argument positioning as a filtering step improves the word-level event field extraction performance from 0.726 to 0.743 macro-F1, outperforming large language models like GPT-4-turbo (with 0.549 F1) and SOTA NER model SoftLexicon (with 0.733 F1). Furthermore, when evaluating the main event extraction task, the proposed model achieves 0.784 macro-F1.

Multi-LoRA continual learning based instruction tuning framework for universal information extraction

Universal information extraction (Universal IE) aims to develop one model capable of solving multiple IE target tasks. Previous works have enhanced extraction performance of target tasks through auxiliary tasks. However, there are still limitations in terms of learning strategies. From one aspect, joint learning-based universal IE approaches, which simply mix auxiliary tasks with target tasks, fail to enable the model to master basic knowledge from auxiliary tasks before learning target tasks. From another aspect, continual learning-based universal IE approaches, which sequentially update all the model parameters on auxiliary tasks and target tasks, tend to cause catastrophic forgetting. In this study, we design a multi-LoRA continual learning-based instruction fine-tuning framework for universal IE. Specifically, we design unique LoRA modules for learning auxiliary tasks and target tasks. We first freeze pre-trained weights and update additional parameters on auxiliary tasks through one LoRA module. Subsequently, we keep the weights frozen and further adjust parameters through another LoRA module to adapt the model to the target tasks. Finally, we merge the frozen weights with learned weights, thereby enabling the model to better leverage the acquired abilities during the inference phase. Therefore, our model masters basic extraction abilities before learning target tasks and does not forget this basic knowledge during the target learning process. Moreover, we regard extraction, classification, and recognition as basic abilities and further design auxiliary tasks based on these basic abilities. Experimental results on 37 datasets across 3 tasks show that our approach reaches state-of-the-art performance.

C3E: A framework for chart classification and content extraction

Incorporating charts into technical documents enhances richness by simplifying complex data representation and improving comprehension. However, automated chart content extraction (CCE) presents a significant challenge within the domain of document analysis and understanding. The CCE problem can be viewed through a series of six sub-tasks: chart classification (CC), text detection and recognition (TDR), text role classification (TRC), axis analysis, legend analysis, and data extraction. Improving these sub-tasks is important for enhancing the effectiveness of CCE. This paper introduces the chart classification and content extraction (C3E) framework, with a primary focus on the first three sub-tasks of CCE: CC, TDR, and TRC. We propose a ChartVision model for the CC, an EfficientNet-based model coupled with a dual-branch architecture incorporating a novel hybrid convolutional and dilated attention module. For text detection and TRC, we introduce a novel CCE method based on YOLOv5, CCE-YOLO, designed for localizing and classifying textual components of varying sizes. Further, for text recognition, we employ a convolutional recurrent neural network with connectionist temporal classification loss. We conducted experimental analysis on benchmark datasets to assess the effectiveness of our approach across each sub-task. Specifically, we evaluated CC, TDR, and TRC methods using the UB-PMC 2020 and UB-PMC 2022 datasets from the ICPR2020 and ICPR2022 CHART-Infographics competitions. The C3E framework achieved notable F1-scores of 94.26%, 92.44%, and 80.64% for CC, TDR, and TRC, respectively on the UB-PMC 2020 dataset and 94.0%, 91.98%, and 84.48% for CC, TDR, and TRC, respectively on the UB-PMC 2022 dataset.

HierbaNetV1: a novel feature extraction framework for deep learning-based weed identification

HierbaNetV1: a novel feature extraction framework for deep learning-based weed identification

MaskDGNets: Masked-attention guided dynamic graph aggregation network for event extraction.

Considering that the traditional deep learning event extraction method ignores the correlation between word features and sequence information, it cannot fully explore the hidden associations between events and events and between events and primary attributes. To solve these problems, we developed a new framework for event extraction called the masked attention-guided dynamic graph aggregation network. On the one hand, to obtain effective word representation and sequence representation, an interaction and complementary relationship are established between word vectors and character vectors. At the same time, a squeeze layer is introduced in the bidirectional independent recurrent unit to model the sentence sequence from both positive and negative directions while retaining the local spatial details to the maximum extent and establishing practical long-term dependencies and rich global context representations. On the other hand, the designed masked attention mechanism can effectively balance the word vector features and sequence semantics and refine these features. The designed dynamic graph aggregation module establishes effective connections between events and events, and between events and essential attributes, strengthens the interactivity and association between them, and realizes feature transfer and aggregation on graph nodes in the neighborhood through dynamic strategies to improve the performance of event extraction. We designed a reconstructed weighted loss function to supervise and adjust each module individually to ensure the optimal feature representation. Finally, the proposed MaskDGNets framework is evaluated on two baseline datasets, DuEE and CCKS2020. It demonstrates its robustness and event extraction performance, with F1 of 81.443% and 87.382%, respectively.

An approach for accurately extracting vehicle trajectory from aerial videos based on computer vision

Vehicle trajectory data holds valuable information for advanced driving development and traffic analysis. While unmanned aerial vehicle (UAV) offer a broader perspective, the detection of small-scale vehicles in video frames still suffers from low accuracy or is even missed. This study proposes a comprehensive technical framework for accurate vehicle trajectory extraction, encompassing six main components: video stabilization, vehicle detection, vehicle tracking, lane marking detection, coordinate transformation, and data denoising. To mitigate video jitter, the SURF and FLANN stabilization algorithms are utilized. An enhanced detector based on You Only Look Once X (YOLOX) is employed for multi-target vehicle detection, incorporating a shallow feature extraction module within the detection head to improve the performance for low-level and small-scale features. Efficient Channel Attention (ECA) modules are integrated before the neck to further boost the expressiveness. Additionally, a sliding window inference method is applied at the input stage to prevent compression of high-resolution video frames. The Savitzky-Golay filter is used for trajectory noise reduction. Verification results demonstrate that the improved YOLOX achieves a mean average precision (mAP) of 88.7 %, an enhancement of 5.6 % over the original model. When compared to advanced YOLOv7 and YOLOv8 models, the proposed method increases mAP@50 by 7.63 % and 1.07 %, respectively. The Mostly Tracked (MT) trajectories metric reaches 98.9 %, and the root-mean-square error of one-sided localization is approximately 0.05 m. These results confirm that the proposed framework is an effective tool for high-accuracy vehicle trajectory data collection in traffic studies. Additionally, a vehicle trajectory dataset has been developed and is publicly accessible at www.cqskyeyex.com.

Machine Learning Based Framework for Lung Cancer Detection and Image Feature Extraction Using VGG16 with PCA on CT-Scan Images

Machine Learning Based Framework for Lung Cancer Detection and Image Feature Extraction Using VGG16 with PCA on CT-Scan Images

Metal-organic frameworks for atmospheric water extraction: Kinetic analysis and stochastic programming under climate variability

Increasing demands for sustainable and distributed freshwater sources drive the exploration of water extraction from ambient air. This study presents a comprehensive computational approach for optimizing unit harvesting cost of the adsorption-based atmospheric water extraction (AWE) systems. There are three objectives (i) assessing the impact of climate variability: utilizing k-means clustering, utilizing climate data in different regions to explore the effects of ambient conditions in dry-hot (California), humid-hot (Florida), and dry-cold (Wyoming) regions, resulting in a preference for harvesting under humid-hot conditions. (ii) performing kinetic analysis: The derived kinetic model connects climate variability to operational time and regeneration temperature, critical process design variables. (iii) assessing adsorption materials: three materials (MIL-100 (Fe), MOF-303, and ZJNU-30) were assessed revealing the impact of variations in maximum capacity and isotherm shape on performance and cost. The optimization algorithm uses a two stage stochastic programming approach to account for climate variability and enables an optimization that balances the capital and operating costs across a range of temperature and humidity conditions.

TR-Net: Token Relation Inspired Table Filling Network for Joint Entity and Relation Extraction

Recently, table filling models have achieved promising performance in jointly extracting relation triplets from complex sentences, leveraging their inherent structural advantage of delineating entities and relations as table cells. Nonetheless, these models predominantly concentrate on the cells corresponding to entity pairs within the predicted tables, neglecting the interrelations among other token pairs. This oversight can potentially lead to the exclusion of essential token information. To address these challenges, we introduce the Token Relation-Inspired Network (TR-Net), a novel framework for the joint extraction of entities and relations. It encompasses a token relation generator that adaptively constructs a token relation table, concentrating on the prominent token cells. Moreover, it also uses a structure-enhanced encoder that integrates the structural and sequential data of sentences via a highway gate mechanism. Our experimental analysis demonstrates that TR-Net delivers considerable enhancements and achieves state-of-the-art performance on four public datasets.

A generalized framework for identification and geometrical feature extraction of circular and polygonal shaped prisms

Abstract This paper introduces a versatile framework crucial for robotic applications such as object manipulation, mobile robot navigation, and pole climbing. It addresses the identification of geometric shapes and dimensions of diverse objects found in varied environments. The proposed method utilizes LiDAR scanning to capture objects from different angles, generating point clouds merged through transformations and superimpositions. After filtering and slicing, intersections are isolated and projected onto a chosen datum plane. The framework employs Non-Linear Least Square fitting via Gauss Newton iterative approach, utilizing pseudo-inverse Jacobian of a hypotrochoid to approximate polygons. The algorithm consecutively fits polygon prisms, determining the best fit with the least norm of error. Results indicate an average least square error of less than 9% for radius fitting and a high f-score for shape identification.

Ambisonics neural speech extraction with directional feature and rotary steering

In scenes with noise and overlapping speakers, directionally extracting audio tracks corresponding to individual speakers is crucial for immersive and interactive spatial audio systems. Although neural networks have been successful in this task, existing steering approaches for adjusting the direction of neural speech extraction mainly target spatial audio directly collected by microphone arrays, while directional speech extraction with Ambisonics spatial audio is less well studied. Therefore, to encode the target directional information as input for the neural network, this paper proposes two Ambisonics directional features based on the spatial feature difference and beamforming principle: the relative harmonic difference and the directional signal enhancement ratio. Using the special property of Ambisonics' rotation transform, a rotary steering pre-processing is also proposed to align the target speaker's direction with a fixed reference by inversely rotating the sound field, thereby simplifying multi-directional extraction to fixed-directional extraction. Finally, we integrate these proposed approaches with the existing temporal-spectral-spatial filtering neural networks to establish a generalized framework for steerable speech extraction and conduct experiments on a simulated Ambisonics dataset containing multiple speakers and noise sources. The experiments show that the proposed approaches outperform existing conditional steering and can be applied to various existing neural network architectures.