[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Sign in to use this feature.

Years

Between: -

Subjects

remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline

Journals

remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline

Article Types

Countries / Regions

remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline

Search Results (839)

Search Parameters:
Keywords = asymmetric information

Order results
Result details
Results per page
Select all
Export citation of selected articles as:
14 pages, 591 KiB  
Article
Asymmetric Information and Credit Rationing in a Model of Search
by Cemil Selcuk
Games 2025, 16(1), 1; https://doi.org/10.3390/g16010001 (registering DOI) - 2 Jan 2025
Abstract
This paper presents a competitive search model focusing on the impact of asymmetric information on credit markets. We show that limited entry by lenders results in endogenous credit rationing, which, in turn, plays a key role in managing adverse selection and prevents the [...] Read more.
This paper presents a competitive search model focusing on the impact of asymmetric information on credit markets. We show that limited entry by lenders results in endogenous credit rationing, which, in turn, plays a key role in managing adverse selection and prevents the credit market from collapsing. Full article
Show Figures

Figure 1

Figure 1
<p>Credit contracts with complete information.</p>
Full article ">Figure 2
<p>Separating equilibrium with credit rationing.</p>
Full article ">Figure 3
<p>Partial shutdown—availability for high types only.</p>
Full article ">
23 pages, 6304 KiB  
Article
Task-Driven Computing Offloading and Resource Allocation Scheme for Maritime Autonomous Surface Ships Under Cloud–Shore–Ship Collaboration Framework
by Supu Xiu, Ying Zhang, Hualong Chen, Yuanqiao Wen and Changshi Xiao
J. Mar. Sci. Eng. 2025, 13(1), 16; https://doi.org/10.3390/jmse13010016 - 26 Dec 2024
Viewed by 421
Abstract
Currently, Maritime Autonomous Surface Ships (MASS) have become one of the most attractive research areas in shipping and academic communities. Based on the ship-to-shore and ship-to-ship communication network, they can exploit diversified and distributed resources such as shore-based facilities and cloud computing centers [...] Read more.
Currently, Maritime Autonomous Surface Ships (MASS) have become one of the most attractive research areas in shipping and academic communities. Based on the ship-to-shore and ship-to-ship communication network, they can exploit diversified and distributed resources such as shore-based facilities and cloud computing centers to execute a variety of ship applications. Due to the increasing number of MASS and asymmetrical distribution of traffic flows, the transportation management must design an efficient cloud–shore–ship collaboration framework and smart resource allocation strategy to improve the performance of the traffic network and provide high-quality applications to the ships. Therefore, we design a cloud–shore–ship collaboration framework, which integrates ship networking and cloud/edge computing and design the respective task collaboration process. It can effectively support the collaborative interaction of distributed resources in the cloud, onshore, and onboard. Based on the global information of the framework, we propose an intelligent resource allocation method based on Q-learning by combining the relevance, QoS characteristics, and priority of ship tasks. Simulation experiments show that our proposed approach can effectively reduce task latency and system energy consumption while supporting the concurrency of scale tasks. Compared with other analogy methods, the proposed algorithm can reduce the task processing delay by at least 15.7% and the task processing energy consumption by 15.4%. Full article
(This article belongs to the Section Ocean Engineering)
Show Figures

Figure 1

Figure 1
<p>The cloud–shore–ship collaboration framework integrating IoS and cloud/edge computing.</p>
Full article ">Figure 2
<p>The basic task processing of cloud–shore–ship collaboration framework.</p>
Full article ">Figure 3
<p>The diagram of cloud–shore–ship collaboration task processing.</p>
Full article ">Figure 4
<p>The tasks unloading process of MASS.</p>
Full article ">Figure 5
<p>(<b>a</b>) The processing flow of serial tasks; (<b>b</b>) The processing flow of ring tasks.</p>
Full article ">Figure 6
<p>The Q-learning based task offloading algorithm for cloud–shore–ship collaboration.</p>
Full article ">Figure 7
<p>The average task delay of the four algorithms with different task numbers.</p>
Full article ">Figure 8
<p>The system energy consumption of the four algorithms with different task numbers.</p>
Full article ">Figure 9
<p>The impact of the number of ships on the task average delay.</p>
Full article ">Figure 10
<p>The impact of the number of ships on the system energy consumption.</p>
Full article ">Figure 11
<p>The impact of task data volume on the task average delay.</p>
Full article ">Figure 12
<p>The impact of the task data volume on the system energy consumption.</p>
Full article ">Figure 13
<p>The impact of the maximum task delay on the task average delay.</p>
Full article ">Figure 14
<p>The impact of the maximum task delay on the system energy consumption.</p>
Full article ">Figure 15
<p>The impact of the maximum communication power on the task average delay.</p>
Full article ">Figure 16
<p>The impact of the maximum communication power on the system energy consumption.</p>
Full article ">
22 pages, 8365 KiB  
Article
FP-YOLOv8: Surface Defect Detection Algorithm for Brake Pipe Ends Based on Improved YOLOv8n
by Ke Rao, Fengxia Zhao and Tianyu Shi
Sensors 2024, 24(24), 8220; https://doi.org/10.3390/s24248220 - 23 Dec 2024
Viewed by 329
Abstract
To address the limitations of existing deep learning-based algorithms in detecting surface defects on brake pipe ends, a novel lightweight detection algorithm, FP-YOLOv8, is proposed. This algorithm is developed based on the YOLOv8n framework with the aim of improving accuracy and model lightweight [...] Read more.
To address the limitations of existing deep learning-based algorithms in detecting surface defects on brake pipe ends, a novel lightweight detection algorithm, FP-YOLOv8, is proposed. This algorithm is developed based on the YOLOv8n framework with the aim of improving accuracy and model lightweight design. First, the C2f_GhostV2 module has been designed to replace the original C2f module. It reduces the model’s parameter count through its unique design. It achieves improved feature representation by adopting specific technique within its structure. Additionally, it incorporates the decoupled fully connected (DFC) attention mechanism, which minimizes information loss during long-range feature transmission by separately capturing pixel information along horizontal and vertical axes via convolution. Second, the Dynamic ATSS label allocation strategy is applied, which dynamically adjusts label assignments by integrating Anchor IoUs and predicted IoUs, effectively reducing the misclassification of high-quality prediction samples as negative samples. Thus, it improves the detection accuracy of the model. Lastly, an asymmetric small-target detection head, FADH, is proposed to utilize depth-separable convolution to accomplish classification and regression tasks, enabling more precise capture of detailed information across scales and improving the detection of small-target defects. The experimental results show that FP-YOLOv8 achieves a mAP50 of 89.5% and an F1-score of 87% on the ends surface defects dataset, representing improvements of 3.3% and 6.0%, respectively, over the YOLOv8n algorithm, Meanwhile, it reduces model parameters and computational costs by 14.3% and 21.0%. Additionally, compared to the baseline model, the AP50 values for cracks, scratches, and flash defects rise by 5.5%, 5.6%, and 2.3%, respectively. These results validate the efficacy of FP-YOLOv8 in enhancing defect detection accuracy, reducing missed detection rates, and decreasing model parameter counts and computational demands, thus meeting the requirements of online defect detection for brake pipe ends surfaces. Full article
(This article belongs to the Section Fault Diagnosis & Sensors)
Show Figures

Figure 1

Figure 1
<p>The structure of FP-YOLOv8.</p>
Full article ">Figure 2
<p>The structure of the Ghostblockv2.</p>
Full article ">Figure 3
<p>The principles of the cheap operations. Cheap operations use point convolution and depth convolution to obtain more feature maps with less computational cost.</p>
Full article ">Figure 4
<p>The principles of the DFC attention mechanism. The horizontal and vertical Fully Connected layers capture the long-range information along the two directions, respectively.</p>
Full article ">Figure 5
<p>Dynamic ATSS network architecture diagram. Dynamic ATSS uses the predicted boxes decoded from the regression branch. The predicted IoUs and anchor IoUs are calculated by comparing the predicted and anchor boxes with the GTs. The Combined IoUs (CIoUs) are obtained by summing the predicted and anchor IoUs. The combined mean and std are calculated similarly. The IoU threshold is the sum of the combined mean and std, and positive candidates are defined as samples with Combined IoUs greater than or equal to the threshold, restricted within the ground truth bounding boxes as final positive samples.</p>
Full article ">Figure 6
<p>The structure of the YOLOv8 detection head and FADH detection head.</p>
Full article ">Figure 7
<p>Brake pipes.</p>
Full article ">Figure 8
<p>Types of surface defects in brake pipe ends.</p>
Full article ">Figure 9
<p>Confusion matrix.</p>
Full article ">Figure 10
<p>Comparison of heatmaps for different algorithms across four types of defects.</p>
Full article ">Figure 11
<p>Score values of each model; (<b>a</b>) mAP50; (<b>b</b>) F1-score.</p>
Full article ">Figure 12
<p>Comparison of each parameter of each model.</p>
Full article ">Figure 13
<p>Illustration of six types of defects in the NEU-DET dataset.</p>
Full article ">Figure 14
<p>Experiments on sensitivity analysis of learning rate and batch size hyperparameters.</p>
Full article ">Figure 15
<p>In-line visual inspection device for brake pipe ends.</p>
Full article ">Figure 16
<p>Actual production environment application. (<b>a</b>) shows the light source and clamp in actual manufacturing, (<b>b</b>) shows the automotive brake pipe being transferred from the clamp to the vision inspection system for defect detection.</p>
Full article ">Figure 17
<p>Brake pipe ends measurement result display interface. The yellow circle represents the outer circle of the brake pipe end and the red circle represents the inner circle of the brake pipe end.</p>
Full article ">
23 pages, 13780 KiB  
Article
Intuitionistic Fuzzy Set Guided Fast Fusion Transformer for Multi-Polarized Petrographic Image of Rock Thin Sections
by Bowei Chen, Bo Yan, Wenqiang Wang, Wenmin He, Yongwei Wang, Lei Peng, Andong Wang and Li Chen
Symmetry 2024, 16(12), 1705; https://doi.org/10.3390/sym16121705 - 23 Dec 2024
Viewed by 369
Abstract
The fusion of multi-polarized petrographic images of rock thin sections involves the fusion of feature information from microscopic images of rock thin sections illuminated under both plane-polarized and orthogonal-polarized light. During the fusion process of rock thin section images, the inherent high resolution [...] Read more.
The fusion of multi-polarized petrographic images of rock thin sections involves the fusion of feature information from microscopic images of rock thin sections illuminated under both plane-polarized and orthogonal-polarized light. During the fusion process of rock thin section images, the inherent high resolution and abundant feature information of the images pose substantial challenges in terms of computational complexity when dealing with massive datasets. In engineering applications, to ensure the quality of image fusion while meeting the practical requirements for high-speed processing, this paper proposes a novel fast fusion Transformer. The model leverages a soft matching algorithm based on intuitionistic fuzzy sets to merge redundant tokens, effectively mitigating the negative effects of asymmetric dependencies between tokens. The newly generated artificial tokens serve as brokers for the Query (Q), forming a novel lightweight fusion strategy. Both subjective visual observations and quantitative analyses demonstrate that the Transformer proposed in this paper is comparable to existing fusion methods in terms of performance while achieving a notable enhancement in its inference efficiency. This is made possible by the attention paradigm, which is equivalent to a generalized form of linear attention, and the newly designed loss function. The model has been experimented on with multiple datasets of different rock types and has exhibited robust generalization capabilities. It provides potential for future research in diverse geological conditions and broader application scenarios. Full article
(This article belongs to the Section Computer)
Show Figures

Figure 1

Figure 1
<p>Thin section images of rocks of different species and polarization modes with a scaling dimension of 500 micrometer.</p>
Full article ">Figure 2
<p>Structure of the proposed fast rock thin sections image fusion broker Transformer.</p>
Full article ">Figure 3
<p>The diagrams on the <b>left</b> and <b>right</b> are respectively the schematic representations of the broker attention module and the linear attention module.</p>
Full article ">Figure 4
<p>The demonstration of the fusion process of various types of rock thin section images. Each row represents a set of rock data, while each column corresponds to an image category. “Pp” and “Op” are abbreviations for “plane-polarized” and “orthogonal polarization”, respectively. The small red circles represent feature markers that have been detected.</p>
Full article ">Figure 5
<p>The fusion results of images of dacitic crystal–lithic–vitric welded tuff by different models. (<b>a</b>) Nestfuse. (<b>b</b>) SEDRFuse. (<b>c</b>) DDcGAN. (<b>d</b>) DenseFuse. (<b>e</b>) DIDFuse. (<b>f</b>) U2Fusion. (<b>g</b>) STDFusion. (<b>h</b>) Our proposed model. The small red boxes are areas of significant difference that have been selected. The larger box is a zoomed-in display of the area, for a clearer comparison of the fusion effect.</p>
Full article ">Figure 6
<p>The fusion results of granite images by different models. (<b>a</b>) Nestfuse. (<b>b</b>) SEDRFuse. (<b>c</b>) DDcGAN. (<b>d</b>) DenseFuse. (<b>e</b>) DIDFuse. (<b>f</b>) U2Fusion. (<b>g</b>) STDFusion. (<b>h</b>) Our proposed model. The small red boxes are areas of significant difference that have been selected. The larger box is a zoomed-in display of the area, for a clearer comparison of the fusion effect.</p>
Full article ">Figure 7
<p>Fusion results for high- and low-resolution images: (<b>a</b>–<b>d</b>) show fused images with a resolution of 480 × 384, while (<b>e</b>–<b>h</b>) show fused images with a resolution of 1280 × 1024.</p>
Full article ">Figure 8
<p>Feature matching results: (<b>a</b>–<b>d</b>) show images with a resolution of 1280 × 1024, while (<b>e</b>–<b>h</b>) represent images with a resolution of 480 × 384. Red lines indicate correctly matched feature pairs.</p>
Full article ">Figure 9
<p>Correlation between the estimated spatial error and the Dice coefficient in three attention mechanisms: (<b>a</b>) Softmax, (<b>b</b>) Linear, and (<b>c</b>) Broker.</p>
Full article ">Figure 10
<p>Comparison of cumulative probability distributions for different loss functions on image fusion performance. The metrics represented by each graph are: (<b>a</b>) MI, (<b>b</b>) PSNR, (<b>c</b>) SF, (<b>d</b>) SSIM, (<b>e</b>) <math display="inline"><semantics> <msup> <mi>Q</mi> <mrow> <mi>A</mi> <mi>B</mi> <mo>/</mo> <mi>F</mi> </mrow> </msup> </semantics></math>, (<b>f</b>) CE, and (<b>g</b>) RMSE.</p>
Full article ">Figure 11
<p>Panels (<b>a</b>–<b>h</b>) represent the CEST MRI images acquired at saturation durations of 17, 25, 33, 52, 60, 68, 76, and 84 min, respectively. Panel (<b>i</b>) shows the output result obtained by fusing this series of images.</p>
Full article ">
22 pages, 7862 KiB  
Article
Vision-Based Deep Reinforcement Learning of Unmanned Aerial Vehicle (UAV) Autonomous Navigation Using Privileged Information
by Junqiao Wang, Zhongliang Yu, Dong Zhou, Jiaqi Shi and Runran Deng
Drones 2024, 8(12), 782; https://doi.org/10.3390/drones8120782 - 22 Dec 2024
Viewed by 331
Abstract
The capability of UAVs for efficient autonomous navigation and obstacle avoidance in complex and unknown environments is critical for applications in agricultural irrigation, disaster relief and logistics. In this paper, we propose the DPRL (Distributed Privileged Reinforcement Learning) navigation algorithm, an end-to-end policy [...] Read more.
The capability of UAVs for efficient autonomous navigation and obstacle avoidance in complex and unknown environments is critical for applications in agricultural irrigation, disaster relief and logistics. In this paper, we propose the DPRL (Distributed Privileged Reinforcement Learning) navigation algorithm, an end-to-end policy designed to address the challenge of high-speed autonomous UAV navigation under partially observable environmental conditions. Our approach combines deep reinforcement learning with privileged learning to overcome the impact of observation data corruption caused by partial observability. We leverage an asymmetric Actor–Critic architecture to provide the agent with privileged information during training, which enhances the model’s perceptual capabilities. Additionally, we present a multi-agent exploration strategy across diverse environments to accelerate experience collection, which in turn expedites model convergence. We conducted extensive simulations across various scenarios, benchmarking our DPRL algorithm against state-of-the-art navigation algorithms. The results consistently demonstrate the superior performance of our algorithm in terms of flight efficiency, robustness and overall success rate. Full article
Show Figures

Figure 1

Figure 1
<p>The DPRL framework for UAV navigation.</p>
Full article ">Figure 2
<p>The effect of adding noise to visual perception data. (<b>a</b>) Depth image obtained from the camera in the simulation environment. (<b>b</b>) Salt-and-pepper noise added to the depth image. (<b>c</b>) Gaussian noise added to image (<b>b</b>). (<b>d</b>) Motion blur applied to image (<b>c</b>).</p>
Full article ">Figure 3
<p>TD3 framework within the POMDP model.</p>
Full article ">Figure 4
<p>Architecture of the Actor and Critic Network.</p>
Full article ">Figure 5
<p>Simulation environment built using UE4 and AirSim. (<b>a</b>) Top-down view of the training environment. (<b>b</b>) Top-down view of Cylindrical Maze, with cylindrical obstacles at random positions and made of random materials. (<b>c</b>) Top-down view of Cubic Maze, with cubic obstacles at random positions and made of random materials. (<b>d</b>) Top-down view of Dense Forest, with trees at random positions and of random types. (<b>e</b>) Top-down view of Mixed Terrain, with obstacles at random positions and of random types. (<b>f</b>) View of the UAV flying within the environment.</p>
Full article ">Figure 6
<p>The training curves of different UAV navigation algorithms. (<b>a</b>) SR curve of Proposed DPRL and TD3. (<b>b</b>) AER curve of Proposed DPRL and TD3.</p>
Full article ">Figure 7
<p>Comparison results of navigation and obstacle avoidance trajectories. Each figure represents the evaluation results of 30 episodes, where blue trajectories indicate successful completions and red trajectories represent failures due to collisions. (<b>a</b>) Flight trajectories of DPRL in training environment. (<b>b</b>) Flight trajectories of TD3 in training environment. (<b>c</b>) Flight trajectories of EGO-Planner-v2 in training environment. (<b>d</b>) Flight trajectories of DPRL in Cylindrical Maze. (<b>e</b>) Flight trajectories of TD3 in Cylindrical Maze. (<b>f</b>) Flight trajectories of EGO-Planner-v2 in Cylindrical Maze.</p>
Full article ">Figure 8
<p>Ablation experiment results for key components of DPRL during model training. (<b>a</b>) SR curves of proposed DPRL, privileged RL and distributed RL. (<b>b</b>) AER curves of proposed DPRL, privileged RL, and distributed RL.</p>
Full article ">Figure 9
<p>Ablation experiment results for state and action space design in DPRL during model training. (<b>a</b>) SR curve of DPRL with 3D and 4D action space. (<b>b</b>) AER curve of DPRL with 3D and 4D action space.</p>
Full article ">
26 pages, 552 KiB  
Article
Sleep Stage Classification Through HRV, Complexity Measures, and Heart Rate Asymmetry Using Generalized Estimating Equations Models
by Bartosz Biczuk, Sebastian Żurek, Szymon Jurga, Elżbieta Turska, Przemysław Guzik and Jarosław Piskorski
Entropy 2024, 26(12), 1100; https://doi.org/10.3390/e26121100 - 16 Dec 2024
Viewed by 430
Abstract
This study investigates whether heart rate asymmetry (HRA) parameters offer insights into sleep stages beyond those provided by conventional heart rate variability (HRV) and complexity measures. Utilizing 31 polysomnographic recordings, we focused exclusively on electrocardiogram (ECG) data, specifically the RR interval time [...] Read more.
This study investigates whether heart rate asymmetry (HRA) parameters offer insights into sleep stages beyond those provided by conventional heart rate variability (HRV) and complexity measures. Utilizing 31 polysomnographic recordings, we focused exclusively on electrocardiogram (ECG) data, specifically the RR interval time series, to explore heart rate dynamics associated with different sleep stages. Employing both statistical techniques and machine learning models, with the Generalized Estimating Equation model as the foundational approach, we assessed the effectiveness of HRA in identifying and differentiating sleep stages and transitions. The models including asymmetric variables for detecting deep sleep stages, N2 and N3, achieved AUCs of 0.85 and 0.89, respectively, those for transitions N2–R, R–N2, i.e., falling in and out of REM sleep, achieved AUCs of 0.85 and 0.80, and those for W–N1, i.e., falling asleep, an AUC of 0.83. All these models were highly statistically significant. The findings demonstrate that HRA parameters provide significant, independent information about sleep stages that is not captured by HRV and complexity measures alone. This additional insight into sleep physiology potentially leads to a better understanding of hearth rhythm during sleep and devising more precise diagnostic tools, including cheap portable devices, for identifying sleep-related disorders. Full article
(This article belongs to the Section Multidisciplinary Applications)
Show Figures

Figure 1

Figure 1
<p>Distribution of sleep stages and transitions within individual subjects.</p>
Full article ">
27 pages, 10096 KiB  
Article
Comparative Analysis of Conventional CNN v’s ImageNet Pretrained ResNet in Medical Image Classification
by Christos Raptis, Efstratios Karavasilis, George Anastasopoulos and Adam Adamopoulos
Information 2024, 15(12), 806; https://doi.org/10.3390/info15120806 - 14 Dec 2024
Viewed by 586
Abstract
Convolutional Neural Networks (CNNs) are the prevalent technology in computer vision and have become increasingly popular for medical imaging data classification and analysis. In this field, due to the scarcity of medical data, pretrained ResNets on ImageNet can be considered a suitable first [...] Read more.
Convolutional Neural Networks (CNNs) are the prevalent technology in computer vision and have become increasingly popular for medical imaging data classification and analysis. In this field, due to the scarcity of medical data, pretrained ResNets on ImageNet can be considered a suitable first approach. This paper examines the medical imaging classification accuracy of conventional basic custom CNNs compared to ImageNet pretrained ResNets on various medical datasets in an effort to give more information about the importance of medical data and its preprocessing techniques for disease studies. Microscope-extracted cytological images were examined along with chest X-rays, MRI brain scans, and melanoma photographs. The medical images were examined in various sets, class combinations, and resolutions. Augmented image datasets and asymmetrical training and validation splits among the classes were also examined. Models were developed after they were tested and fine-tuned with respect to their network size, parameter values and network methods, image resolution, size of dataset, multitude, and genre of class. Overfitting was also examined, and comparative studies regarding the computational cost of different models were performed. The models achieved high accuracy in image classification that varies depending on the dataset and can be easily incorporated in future over-the-internet medical decision-supporting (telemedicine) environments. In addition, it appeared that conventional basic custom CNN overperformed ImageNet pretrained ResNets. The obtained results indicate the importance of utilizing medical image data as a testbed for improvements in CNN classification performance and the possibility of using CNNs and data preprocessing techniques for disease studies. Full article
Show Figures

Figure 1

Figure 1
<p>Pap smear test: (<b>a</b>) koilocytotic cells (original size); (<b>b</b>) koilocytotic cells (zoomed and cropped). Images from Axioscope 5 microscope camera Axiocam 208 Color.</p>
Full article ">Figure 2
<p>Adult chest X-rays. From top to bottom and left to right: 1,3.5 viral pneumonia; 2,4 bacterial pneumonia. Images from the adult chest X-ray dataset that was used to train the models.</p>
Full article ">Figure 3
<p>From top to bottom and left to right: original image; brighter; darker; horizontal flip; rotated; vertical flip (Glioma). Glioma tumor image from the dataset used to train the models.</p>
Full article ">Figure 4
<p>(<b>a</b>) Malignant melanoma; (<b>b</b>) benign melanoma. Melanoma photographs from the dataset used to train the models.</p>
Full article ">Figure 5
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training error (blue) of the melanoma classification models: (<b>a</b>) 15-layer conventional CNN model trained on melanoma photographs; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 5 Cont.
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training error (blue) of the melanoma classification models: (<b>a</b>) 15-layer conventional CNN model trained on melanoma photographs; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 6
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training error (blue) of the adult chest X-ray classification models: (<b>a</b>) 15-layer conventional CNN model trained on chest X-rays; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 6 Cont.
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training error (blue) of the adult chest X-ray classification models: (<b>a</b>) 15-layer conventional CNN model trained on chest X-rays; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 7
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training error (blue) of the pediatric X-ray classification models: (<b>a</b>) 17-layer conventional CNN model trained on pediatric chest X-rays; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 7 Cont.
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training error (blue) of the pediatric X-ray classification models: (<b>a</b>) 17-layer conventional CNN model trained on pediatric chest X-rays; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 8
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training error (blue) of the MRI brain scans classification models: (<b>a</b>) 19-layer conventional CNN model trained on MRI brain scans; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 8 Cont.
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training error (blue) of the MRI brain scans classification models: (<b>a</b>) 19-layer conventional CNN model trained on MRI brain scans; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 9
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training (blue) error of the pap smear test classification models: (<b>a</b>) 17-layer conventional CNN model trained on pap smear tests; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 9 Cont.
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training (blue) error of the pap smear test classification models: (<b>a</b>) 17-layer conventional CNN model trained on pap smear tests; (<b>b</b>) ResNet50 ImageNet pretrained model; (<b>c</b>) MobileNetV2 ImageNet pretrained model.</p>
Full article ">Figure 10
<p>Validation (orange)/training accuracy (blue) and validation (orange)/training (blue) error of the fine-tuned ResNet50 ImageNet pretrained model for the adult Chest X-rays dataset. The green line shows the point (epoch 20) when fine-tuning starts.</p>
Full article ">Figure A1
<p>ResNet50 model architecture (residual network).</p>
Full article ">Figure A2
<p>MobileNet model architecture (inverted residual network).</p>
Full article ">Figure A3
<p>MobileNetV1 and MobileNetV2.</p>
Full article ">Figure A4
<p>Identity block.</p>
Full article ">Figure A5
<p>Convolutional block.</p>
Full article ">Figure A6
<p>Residual network connection.</p>
Full article ">
16 pages, 1428 KiB  
Article
A Definition of a Heywood Case in Item Response Theory Based on Fisher Information
by Jay Verkuilen and Peter J. Johnson
Entropy 2024, 26(12), 1096; https://doi.org/10.3390/e26121096 - 14 Dec 2024
Viewed by 428
Abstract
Heywood cases and other improper solutions occur frequently in latent variable models, e.g., factor analysis, item response theory, latent class analysis, multilevel models, or structural equation models, all of which are models with response variables taken from an exponential family. They have important [...] Read more.
Heywood cases and other improper solutions occur frequently in latent variable models, e.g., factor analysis, item response theory, latent class analysis, multilevel models, or structural equation models, all of which are models with response variables taken from an exponential family. They have important consequences for scoring with the latent variable model and are indicative of issues in a model, such as poor identification or model misspecification. In the context of the 2PL and 3PL models in IRT, they are more frequently known as Guttman items and are identified by having a discrimination parameter that is deemed excessively large. Other IRT models, such as the newer asymmetric item response theory (AsymIRT) or polytomous IRT models often have parameters that are not easy to interpret directly, so scanning parameter estimates are not necessarily indicative of the presence of problematic values. The graphical examination of the IRF can be useful but is necessarily subjective and highly dependent on choices of graphical defaults. We propose using the derivatives of the IRF, item Fisher information functions, and our proposed Item Fraction of Total Information (IFTI) decomposition metric to bypass the parameters, allowing for the more concrete and consistent identification of Heywood cases. We illustrate the approach by using empirical examples by using AsymIRT and nominal response models. Full article
(This article belongs to the Special Issue Applications of Fisher Information in Sciences II)
Show Figures

Figure 1

Figure 1
<p>(<b>a</b>) IRFs and (<b>b</b>) IIFs for a typical, suspect, and problematic item. Note the IIFs are put on a <math display="inline"><semantics> <msqrt> <mrow> <mi>I</mi> <mo>(</mo> <mi>θ</mi> <mo>)</mo> </mrow> </msqrt> </semantics></math> scale for ease of visualization.</p>
Full article ">Figure 2
<p>(<b>a</b>) Fitted <math display="inline"><semantics> <mrow> <mi>π</mi> <mo>(</mo> <mi>θ</mi> <mo>)</mo> </mrow> </semantics></math> and (<b>b</b>) <math display="inline"><semantics> <mrow> <msub> <mo>∂</mo> <mi>θ</mi> </msub> <mi>π</mi> <mrow> <mo>(</mo> <mi>θ</mi> <mo>)</mo> </mrow> </mrow> </semantics></math> for Example 1. The mental rotation items are colored; item MR3 is orange, item MR4 is green, item MR6 is cyan, and item MR8 is red.</p>
Full article ">Figure 3
<p>(<b>a</b>) Test information function and (<b>b</b>) fitted <math display="inline"><semantics> <mrow> <mi>I</mi> <mo>(</mo> <mi>θ</mi> <mo>)</mo> </mrow> </semantics></math> for Example 1. The mental rotation items are colored; item MR3 is orange, item MR4 is green, item MR6 is cyan, and item MR8 is red. Note that the overall trend in the item information plots is reflected in the test information function.</p>
Full article ">Figure 4
<p>(<b>a</b>) Fitted <math display="inline"><semantics> <mrow> <mi>I</mi> <mi>F</mi> <mi>T</mi> <mi>I</mi> </mrow> </semantics></math> for the total 16 items and (<b>b</b>) fitted <math display="inline"><semantics> <mrow> <mi>I</mi> <mi>F</mi> <mi>T</mi> <mi>I</mi> </mrow> </semantics></math> for Example 1 with the two most severe Heywood case items removed. The mental rotation items are colored; item MR3 is orange, item MR4 is green, item MR6 is cyan, and item MR8 is red.</p>
Full article ">Figure 5
<p>Box plots of (<b>a</b>) EAP predicted scores (<math display="inline"><semantics> <mover accent="true"> <mi>θ</mi> <mo>^</mo> </mover> </semantics></math>) and (<b>b</b>) standard errors (<math display="inline"><semantics> <mrow> <mi>SE</mi> <mo>(</mo> <mover accent="true"> <mi>θ</mi> <mo>^</mo> </mover> <mo>)</mo> </mrow> </semantics></math>), shown over the proportion correct. Note that while most boxes are fairly modest, for high proportions correct, the boxes are unexpectedly wide, indicating the instability induced by the mental rotation items.</p>
Full article ">Figure 6
<p>(<b>a</b>) Fitted <math display="inline"><semantics> <mrow> <mi>I</mi> <mi>F</mi> <mi>T</mi> <mi>I</mi> </mrow> </semantics></math> and (<b>b</b>) box plots of EAP predicted scores for Example 1, with a loose N(0,1) prior set on the asymmetry parameter of the RH model. The mental rotation items are colored; item MR3 is orange, item MR4 is green, item MR6 is cyan, and item MR8 is red.</p>
Full article ">Figure 7
<p>(<b>a</b>) Fitted <math display="inline"><semantics> <mrow> <mi>I</mi> <mi>F</mi> <mi>T</mi> <mi>I</mi> </mrow> </semantics></math> and (<b>b</b>) box plots of EAP predicted scores for Example 1, with a strict N(0,0.25) prior set on the asymmetry parameter of the RH model. The mental rotation items are colored; item MR3 is orange, item MR4 is green, item MR6 is cyan, and item MR8 is red.</p>
Full article ">Figure 8
<p>(<b>a</b>) Fitted <math display="inline"><semantics> <mrow> <mi>I</mi> <mo>(</mo> <mi>θ</mi> <mo>)</mo> </mrow> </semantics></math> and (<b>b</b>) fitted <math display="inline"><semantics> <mrow> <mi>I</mi> <mi>F</mi> <mi>T</mi> <mi>I</mi> <mo>(</mo> <mi>θ</mi> <mo>)</mo> </mrow> </semantics></math> for Example 2. Item 11 is bolded.</p>
Full article ">Figure 9
<p>(<b>a</b>) Fitted <math display="inline"><semantics> <mrow> <mi>I</mi> <mi>F</mi> <mi>T</mi> <mi>I</mi> <mo>(</mo> <mi>θ</mi> <mo>)</mo> </mrow> </semantics></math> without item 11 and (<b>b</b>) fitted <math display="inline"><semantics> <mrow> <mi>I</mi> <mi>F</mi> <mi>T</mi> <mi>I</mi> <mo>(</mo> <mi>θ</mi> <mo>)</mo> </mrow> </semantics></math> without items 17–32 for Example 2. Items 31 and 11 are bold in <b>a</b> and <b>b</b>, respectively.</p>
Full article ">
31 pages, 10015 KiB  
Review
The Enantiopure 1,2-Diphenylethylenediamine (DPEDA) Motif in the Development of Organocatalysts for Asymmetric Reactions: Advances in the Last 20 Years
by Shilashi Badasa Oljira, Martina De Angelis, Andrea Sorato, Giulia Mazzoccanti, Simone Manetto, Ilaria D’Acquarica and Alessia Ciogli
Catalysts 2024, 14(12), 915; https://doi.org/10.3390/catal14120915 - 12 Dec 2024
Viewed by 709
Abstract
1,2-Diphenylethylenediamine (DPEDA) is a privileged chiral scaffold being used in the construction of a broad variety of organocatalysts and ligands for enantioselective organic reactions. This molecule gave a significant contribution in the synthesis of structurally different bi/multifunctional organocatalysts. DPEDA played an essential role [...] Read more.
1,2-Diphenylethylenediamine (DPEDA) is a privileged chiral scaffold being used in the construction of a broad variety of organocatalysts and ligands for enantioselective organic reactions. This molecule gave a significant contribution in the synthesis of structurally different bi/multifunctional organocatalysts. DPEDA played an essential role in the development of organocatalysts capable of yielding important information on the different reaction mechanisms, like enamine, iminium, hydrogen-bonding and anion-binding catalysis. The aim of the present review is to highlight and summarize the achievements reached in the last 20 years (2004–2024) in the chemistry of DPEDA-based organocatalysts for asymmetric synthesis. Full article
(This article belongs to the Section Catalysis in Organic and Polymer Chemistry)
Show Figures

Graphical abstract

Graphical abstract
Full article ">Figure 1
<p>Chemical structure of the chiral organocatalyst 1,2-diphenylethylendiamine (DPEDA). C<sub>14</sub>H<sub>16</sub>N<sub>2</sub> (MW 212.29). Alternate names: 1,2-diphenyl-1,2-ethanediamine; 1,2-diamino-1,2-diphenylethane; stilbenediamine; α,β-diaminodihydrostilbene [<a href="#B11-catalysts-14-00915" class="html-bibr">11</a>].</p>
Full article ">Figure 2
<p>Monofunctionalization of the DPEDA catalyst.</p>
Full article ">Figure 3
<p>Monofunctional thiourea derivatives of the DPEDA catalyst.</p>
Full article ">Figure 4
<p>Plausible bifunctional mode of action of the primary DPEDA thiourea catalyst with R = 3,5-(CF<sub>3</sub>)<sub>2</sub>-C<sub>6</sub>H<sub>3</sub>.</p>
Full article ">Figure 5
<p>Structures of <span class="html-italic">C</span><sub>2</sub>-symmetric (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA-1,3-bis-thiourea organocatalysts.</p>
Full article ">Figure 6
<p>Structures of primary-tertiary (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA-based (<b>27</b>–<b>33</b>) and (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA-based (<b>34</b>) organocatalysts.</p>
Full article ">Figure 7
<p>General structure of bifunctional squaramides catalysts.</p>
Full article ">Figure 8
<p>Structures of primary-amine squaramide (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA- and (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA-based organocatalysts.</p>
Full article ">Figure 9
<p>Stereochemical model for the Michael reaction of 4-hydroxycoumarin (red) with (<span class="html-italic">E</span>)-4-phenyl-3-buten-2-one (blue) catalyzed by catalyst (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-<b>40</b>.</p>
Full article ">Figure 10
<p>Bifunctionalization of the DPEDA catalyst.</p>
Full article ">Figure 11
<p>Structures of imidazoline DPEDA-based organocatalysts. Each catalyst was obtained as a 1:1 mixture of diastereomers.</p>
Full article ">Figure 12
<p>Multifunctional (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA-based organocatalyst.</p>
Full article ">Figure 13
<p>Structures of bis(amido)-DPEDA-based organocatalysts.</p>
Full article ">Figure 14
<p>Structures of bis-prolinamido- (<b>56</b>,<b>57</b>) and bis-siloxyserinamido- (<b>58</b>,<b>59</b>) DPEDA-based organocatalysts.</p>
Full article ">Figure 15
<p>Bifunctional amino-thiourea derivatives of the (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA catalyst.</p>
Full article ">Figure 16
<p>Bifunctional 3-pentylamino-thiourea derivatives of the DPEDA catalyst.</p>
Full article ">Figure 17
<p>Prolinamide-thiourea derivatives of the (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA catalyst.</p>
Full article ">Figure 18
<p>Sulfonamido-thiourea derivatives of the DPEDA catalyst.</p>
Full article ">Figure 19
<p><span class="html-italic">Cinchona</span> alkaloid-derived bifunctional sulfonamide-thiourea derivative of the (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA catalyst.</p>
Full article ">Figure 20
<p>Structures of tetraamino-bis-thiourea-DPEDA-based cyclic organocatalysts.</p>
Full article ">Figure 21
<p>Structures of salophen-H ligands (<b>81</b>, <b>82</b>) and Acen-H catalyst <b>83</b> based on (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA.</p>
Full article ">Figure 22
<p>Structures of bis-squaramide-type DPEDA-based organocatalysts.</p>
Full article ">Figure 23
<p>Structures of <span class="html-italic">C</span><sub>3</sub>-symmetric trisimidazoline (<b>left</b>) and <span class="html-italic">C</span><sub>2</sub>-symmetric bisimidazoline (<b>right</b>) derivatives of (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA.</p>
Full article ">Figure 24
<p>Structures of monocyclic (<b>left</b>) and bicyclic (<b>right</b>) guanidine catalysts incorporating (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA.</p>
Full article ">Figure 25
<p>Structures of acyclic guanidine DPEDA-based organocatalysts. Cy = cyclohexyl. Ar = 4-MeC<sub>6</sub>H<sub>4</sub>; 2,6-F<sub>2</sub>C<sub>6</sub>H<sub>3</sub>.</p>
Full article ">Figure 26
<p>Structure of Type 1-iminophosphorane derivatives of (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA.</p>
Full article ">Figure 27
<p>Structure of Type 2-iminophosphorane derivatives of (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA.</p>
Full article ">Scheme 1
<p>Enantioselective Diels–Alder cycloaddition between cyclopentadiene and (<span class="html-italic">E</span>)-crotonaldehyde catalyzed by bisammonium salts of (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA.</p>
Full article ">Scheme 2
<p>Synthesis of (<span class="html-italic">R</span>)- or (<span class="html-italic">S</span>)-warfarin <b>7</b> in the presence of DPEDA as the catalyst.</p>
Full article ">Scheme 3
<p>Aziridination of cyclic enones in the presence of (<span class="html-italic">R</span>,<span class="html-italic">R</span>)-DPEDA as the catalyst.</p>
Full article ">Scheme 4
<p>Asymmetric cyclopropanation of cinnamone (R<sup>1</sup> = Ph) with a stabilized sulfur ylide catalyzed by (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA.</p>
Full article ">Scheme 5
<p>Enantioselective synthesis of 3,4-dihydropyran derivatives (R<sup>1</sup> = Ph).</p>
Full article ">Scheme 6
<p>Enantioselective vinylogous aldol reaction with (<span class="html-italic">R</span>,<span class="html-italic">R</span>)-DPEDA-thiourea catalysts.</p>
Full article ">Scheme 7
<p>Asymmetric Michael addition of β-ketoesters to <span class="html-italic">trans</span>-β-nitrostyrene.</p>
Full article ">Scheme 8
<p>Asymmetric cross-aldol reactions of trifluoromethyl aromatic ketones with linear aliphatic ketones using primary-tertiary (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA-based organocatalysts.</p>
Full article ">Scheme 9
<p>Asymmetric Mannich reaction of trifluoromethyl ketimines (X = F, Cl, Br, CF<sub>3</sub>, Me, Ph, <span class="html-italic">t</span>-Bu, OMe) with acetone.</p>
Full article ">Scheme 10
<p>Enantioselective addition of glyoxylate cyanohydrins to <span class="html-italic">N</span>-Boc-imines catalyzed by a benzothiadiazine DPEDA bifunctional derivative.</p>
Full article ">Scheme 11
<p>Asymmetric Mannich reaction of <span class="html-italic">N</span>-phosphinoylimines with malononitrile catalyzed by the mixed-amino-amido-functionalized (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA-based organocatalyst.</p>
Full article ">Scheme 12
<p>Asymmetric Strecker reaction of <span class="html-italic">N</span>-benzhydrylimines with ethyl cyanoformate catalyzed by homochiral and heterochiral bis(amido)-DPEDA-based organocatalysts.</p>
Full article ">Scheme 13
<p>Stereoselective synthesis of 1-deoxy-<span class="html-small-caps">d</span>-ketohexoses catalyzed by bis-prolinamido- and bis-siloxyserinamido-DPEDA-based organocatalysts.</p>
Full article ">Scheme 14
<p>Asymmetric Michael addition of aromatic cyanoacetates to phenyl vinyl sulfone.</p>
Full article ">Scheme 15
<p>Asymmetric ring-opening reaction of cyclic anhydrides using a bifunctional amino-thiourea (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA-based catalyst.</p>
Full article ">Scheme 16
<p>Applications of sulfonamido-thiourea-DPEDA-based catalysts. Sulfa-Michael reactions of dienone systems (<b>A</b>) and nitroolefin enoates (<b>B</b>) to thiols; (<b>C</b>) Michael/Henry cascade reaction of spiranic tricyclic compounds and nitroolefins.</p>
Full article ">Scheme 17
<p>Conjugate additions of nitroalkanes to α,β-unsaturated ketones catalyzed by sulfonamido-thiourea-DPEDA-based catalysts.</p>
Full article ">Scheme 18
<p>Applications of <span class="html-italic">Cinchona</span> alkaloid-derived sulfonamido-thiourea-DPEDA-based catalysts. (<b>A</b>) Conjugate additions of diketones to nitroolefins; (<b>B</b>) Mannich-type reaction of pyrazole amides and protected ketimines.</p>
Full article ">Scheme 19
<p>Applications of tetraamino-bis-thiourea-DPEDA-based cyclic organocatalysts. Decarboxylative Mannich addition of cyclic sulfamate-aldimines and benzoyl acetic acid (<b>A</b>) and malonic acid half thioesters to isatin-derived ketimines (<b>B</b>).</p>
Full article ">Scheme 20
<p>Ring-opening addition of CO<sub>2</sub> to epoxides catalyzed by Acen-H catalysts based on (1<span class="html-italic">R</span>,2<span class="html-italic">R</span>)-DPEDA.</p>
Full article ">Scheme 21
<p>Asymmetric Michael addition of β-diketones to <span class="html-italic">trans</span>-β-nitrostyrene catalyzed by bis-squaramide-type DPEDA-based organocatalysts.</p>
Full article ">Scheme 22
<p>Asymmetric Michael addition of β-ketoesters to <span class="html-italic">trans</span>-β-nitrostyrene catalyzed by imidazoline derivatives of (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA.</p>
Full article ">Scheme 23
<p>Asymmetric bromolactonization of 5-substituted 5-hexenoic acids catalyzed by a <span class="html-italic">C</span><sub>3</sub>-symmetric chiral trisimidazoline derivative of (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA.</p>
Full article ">Scheme 24
<p>Applications of a monocyclic guanidine catalyst incorporating (1<span class="html-italic">S</span>,2<span class="html-italic">S</span>)-DPEDA. (<b>A</b>) Mannich reaction between diphenyliminoglycinate and ethyl acrylate; (<b>B</b>) Imine-imine cross-coupling reaction between glyoxylate imine and <span class="html-italic">N</span>-Boc imines.</p>
Full article ">Scheme 25
<p>Guanidinium-catalyzed phospha-Mannich reaction of secondary phosphine oxides with imines.</p>
Full article ">Scheme 26
<p>Oxyamination of azlactones and simultaneous kinetic resolution of oxaziridines catalyzed by DPEDA-based bisguanidine catalysts.</p>
Full article ">Scheme 27
<p>Electrophilic amination of 2-alkyltetralones with azodicarboxylates catalyzed by Type 1-DPEDA-based iminophosphoranes (13 salts of catalysts were tested).</p>
Full article ">Scheme 28
<p>Nitro-Mannich reaction of nitromethane with <span class="html-italic">N</span>-diphenylphosphinoyl ketimines catalyzed by Type 2-DPEDA-based iminophosphoranes.</p>
Full article ">
18 pages, 361 KiB  
Article
More Quality, Less Trust?
by Michael Dreyfuss, Yahel Giat and Eran Manes
Int. J. Financial Stud. 2024, 12(4), 123; https://doi.org/10.3390/ijfs12040123 - 9 Dec 2024
Viewed by 428
Abstract
This study investigates how an increase in the quality of business ventures, measured as their success probability, affects trust and return on investment (ROI) in situations where the investor–entrepreneur interaction is affected by moral hazard and asymmetric information. We model a repeated trust [...] Read more.
This study investigates how an increase in the quality of business ventures, measured as their success probability, affects trust and return on investment (ROI) in situations where the investor–entrepreneur interaction is affected by moral hazard and asymmetric information. We model a repeated trust problem between investors and entrepreneurs, featuring moral hazard and adverse selection. Hidden Markov techniques and computer simulations are used to derive the main results. We find that trust and ROI may decline as quality improves. Although lenders tend to reduce the requirements for granting initial credit, they nevertheless become less tolerant of current borrowers who fail to pay back. Additionally, we demonstrate a novel substitution effect, where lenders prefer new borrowers over existing borrowers that experienced early failures. The main conclusions of our study are that while impressing early on is effective in gaining first access to credit, it may nevertheless hurt the cause of getting credit in subsequent periods, following an early failure. In business environments plagued with ex post moral hazard, entrepreneurs might do better by gaining trust first and impressing later. Furthermore, our results imply that in a thriving economy, not only are bad loans made, but good loans are lost as well. Full article
Show Figures

Figure 1

Figure 1
<p>The Markov chain modeling of the repeated trust problem. Notes: Each node denotes a different state. State 0 is the beginning state. All the states transition to it with probability <span class="html-italic">d</span>. The numeric states <span class="html-italic">i</span>, <math display="inline"><semantics> <mrow> <mn>0</mn> <mo>≤</mo> <mi>i</mi> <mo>≤</mo> <mi>m</mi> </mrow> </semantics></math>, denote states in which paying back never occurred. The transition probability between state <span class="html-italic">i</span> to <math display="inline"><semantics> <mrow> <mi>i</mi> <mo>+</mo> <mn>1</mn> </mrow> </semantics></math> (for <math display="inline"><semantics> <mrow> <mn>0</mn> <mo>≤</mo> <mi>i</mi> <mo>&lt;</mo> <mi>m</mi> </mrow> </semantics></math>) is <math display="inline"><semantics> <mrow> <mrow> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>d</mi> <mo>)</mo> </mrow> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>q</mi> <mrow> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>t</mi> <mo>)</mo> </mrow> <msub> <mover accent="true"> <mi>p</mi> <mo stretchy="false">^</mo> </mover> <mi>i</mi> </msub> <mo>)</mo> </mrow> </semantics></math>. State <span class="html-italic">m</span> is when the investor distrusts the entrepreneur and transitions to itself with probability <math display="inline"><semantics> <mrow> <mn>1</mn> <mo>−</mo> <mi>d</mi> </mrow> </semantics></math>. State <span class="html-italic">Y</span> is when the last period ended with paying back. All the numeric states transition to it with probability <math display="inline"><semantics> <mrow> <mrow> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>d</mi> <mo>)</mo> </mrow> <mi>q</mi> <mrow> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>t</mi> <mo>)</mo> </mrow> <msub> <mover accent="true"> <mi>p</mi> <mo stretchy="false">^</mo> </mover> <mi>i</mi> </msub> </mrow> </semantics></math>. State <span class="html-italic">N</span> is when there was paying back in the past but not in the last period. States <span class="html-italic">Y</span> and <span class="html-italic">N</span> transition into <span class="html-italic">Y</span> with probability <math display="inline"><semantics> <mrow> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>d</mi> <mo>)</mo> <mi>q</mi> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>t</mi> <mo>)</mo> </mrow> </semantics></math> and into <span class="html-italic">N</span> with probability <math display="inline"><semantics> <mrow> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>d</mi> <mo>)</mo> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>q</mi> <mo>(</mo> <mn>1</mn> <mo>−</mo> <mi>t</mi> <mo>)</mo> <mo>)</mo> </mrow> </semantics></math>.</p>
Full article ">Figure 2
<p>Trust panel (<b>a</b>) and ROI panel (<b>b</b>) as a function of the venture’s success probability, <span class="html-italic">q</span>. Notes: the values used for these graphs are <math display="inline"><semantics> <mrow> <mi>d</mi> <mo>=</mo> <mn>10</mn> <mo>%</mo> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>ρ</mi> <mo>=</mo> <mn>9</mn> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>r</mi> <mo>=</mo> <mn>5</mn> <mo>%</mo> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <mi>p</mi> <mo>=</mo> <mn>50</mn> <mo>%</mo> </mrow> </semantics></math>.</p>
Full article ">Figure 3
<p>Trust as a function of the return on a successful venture, <math display="inline"><semantics> <mi>ρ</mi> </semantics></math>, and the probability of honesty, <span class="html-italic">p</span>. Notes: the values used for these graphs are <math display="inline"><semantics> <mrow> <mi>d</mi> <mo>=</mo> <mn>10</mn> <mo>%</mo> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>r</mi> <mo>=</mo> <mn>5</mn> <mo>%</mo> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>q</mi> <mo>=</mo> <mn>45</mn> <mo>%</mo> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>p</mi> <mo>=</mo> <mn>50</mn> <mo>%</mo> </mrow> </semantics></math> (<b>a</b>) and <math display="inline"><semantics> <mrow> <mi>ρ</mi> <mo>=</mo> <mn>9</mn> </mrow> </semantics></math> (<b>b</b>).</p>
Full article ">Figure 4
<p>Trust as a function of the project’s quality for different degrees of honesty. Notes: the values used for these graphs are <math display="inline"><semantics> <mrow> <mi>d</mi> <mo>=</mo> <mn>10</mn> <mo>%</mo> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>ρ</mi> <mo>=</mo> <mn>9</mn> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>r</mi> <mo>=</mo> <mn>5</mn> <mo>%</mo> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <mi>t</mi> <mo>=</mo> <mn>10</mn> <mo>%</mo> </mrow> </semantics></math>.</p>
Full article ">Figure 5
<p>The <span class="html-italic">q</span> that maximizes trust panel (<b>a</b>) and ROI panel (<b>b</b>) as a function of the probability of honesty, <span class="html-italic">p</span>. Notes: the values used for these graphs are <math display="inline"><semantics> <mrow> <mi>d</mi> <mo>=</mo> <mn>10</mn> <mo>%</mo> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>r</mi> <mo>=</mo> <mn>5</mn> <mo>%</mo> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>t</mi> <mo>=</mo> <mn>10</mn> <mo>%</mo> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <mi>ρ</mi> <mo>=</mo> <mn>9</mn> </mrow> </semantics></math>.</p>
Full article ">
20 pages, 1332 KiB  
Article
Differential Transform Method (DTM) and Physics-Informed Neural Networks (PINNs) in Solving Integral–Algebraic Equation Systems
by Rafał Brociek and Mariusz Pleszczyński
Symmetry 2024, 16(12), 1619; https://doi.org/10.3390/sym16121619 - 6 Dec 2024
Viewed by 504
Abstract
Integral–algebraic equations and their systems are a common description of many technical and engineering problems. Often, such models also describe certain dependencies occurring in nature (e.g., ecosystem behaviors). The integral equations occurring in this problem may have two types of domains—symmetric or asymmetric. [...] Read more.
Integral–algebraic equations and their systems are a common description of many technical and engineering problems. Often, such models also describe certain dependencies occurring in nature (e.g., ecosystem behaviors). The integral equations occurring in this problem may have two types of domains—symmetric or asymmetric. Depending on whether such symmetry exists in the system describing a given problem, we must choose the appropriate method to solve this system. In this task, the absence of symmetry is more advantageous, but the presented examples demonstrate how one can approach cases where symmetry is present. In this paper, we present the application of two methods for solving such tasks: the analytical Differential Transform Method (DTM) and Physics-informed Neural Networks (PINNs). We consider a wide class of these types of equation systems, including Volterra and Fredholm integrals (which are also in a single model). We demonstrate that despite the complex nature of the problem, both methods are capable of handling such tasks, and thus, they can be successfully applied to the issues discussed in this article. Full article
(This article belongs to the Special Issue Symmetry in Numerical Analysis and Applied Mathematics)
Show Figures

Figure 1

Figure 1
<p>Schema of physics-informed neural network for solving system of IAEs with boundary conditions.</p>
Full article ">Figure 2
<p>The exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>1</mn> </msub> </semantics></math> (solid green line), the approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>1</mn> <mo>,</mo> <mn>8</mn> </mrow> </msub> </semantics></math> (dashed red line), and the absolute errors <math display="inline"><semantics> <mo>Δ</mo> </semantics></math> of this approximation, as well as the exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>2</mn> </msub> </semantics></math> (solid blue line), the approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>2</mn> <mo>,</mo> <mn>8</mn> </mrow> </msub> </semantics></math> (dashed red line), and absolute errors <math display="inline"><semantics> <mo>Δ</mo> </semantics></math> of this approximation.</p>
Full article ">Figure 3
<p>The exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>1</mn> </msub> </semantics></math> (solid green line), the approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>1</mn> <mo>,</mo> <mn>15</mn> </mrow> </msub> </semantics></math> (dashed red line), and the absolute errors <math display="inline"><semantics> <mo>Δ</mo> </semantics></math> of this approximation, as well as the exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>2</mn> </msub> </semantics></math> (solid blue line), the approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>2</mn> <mo>,</mo> <mn>15</mn> </mrow> </msub> </semantics></math> (dashed red line), and the absolute errors <math display="inline"><semantics> <mo>Δ</mo> </semantics></math> of this approximation.</p>
Full article ">Figure 4
<p>The exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>1</mn> </msub> </semantics></math> (solid green line), the approximation (dashed red line), the absolute errors of approximation of <math display="inline"><semantics> <msub> <mi>y</mi> <mn>1</mn> </msub> </semantics></math> (green dotted line—right side), the exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>2</mn> </msub> </semantics></math> (solid blue line), the approximation (dashed red line), and the absolute errors of approximation of <math display="inline"><semantics> <msub> <mi>y</mi> <mn>2</mn> </msub> </semantics></math> (blue dotted line—right side).</p>
Full article ">Figure 5
<p>Exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>1</mn> </msub> </semantics></math> (solid green line), approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>1</mn> <mo>,</mo> <mn>8</mn> </mrow> </msub> </semantics></math> (dashed red line), exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>2</mn> </msub> </semantics></math> (solid blue line), approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>2</mn> <mo>,</mo> <mn>8</mn> </mrow> </msub> </semantics></math> (dashed red line), exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>3</mn> </msub> </semantics></math> (solid cyan line), approximate solution <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>3</mn> <mo>,</mo> <mn>8</mn> </mrow> </msub> </semantics></math> (dashed red line), and absolute errors <math display="inline"><semantics> <mo>Δ</mo> </semantics></math> of these approximations.</p>
Full article ">Figure 6
<p>Exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mi>i</mi> </msub> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mn>3</mn> </mrow> </semantics></math> (green, blue, and orange solid lines—<b>left side</b>), approximation <math display="inline"><semantics> <msub> <mi>y</mi> <mi>i</mi> </msub> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mn>3</mn> </mrow> </semantics></math> (red dashed line—<b>left side</b>), and absolute errors of these approximations (green, blue, and orange dotted lines—<b>right side</b>) for example 2.</p>
Full article ">Figure 7
<p>Exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>1</mn> </msub> </semantics></math> (solid green line), approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>1</mn> <mo>,</mo> <mn>12</mn> </mrow> </msub> </semantics></math> (dashed red line), exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>2</mn> </msub> </semantics></math> (solid blue line), approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>2</mn> <mo>,</mo> <mn>12</mn> </mrow> </msub> </semantics></math> (dashed red line), exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>3</mn> </msub> </semantics></math> (solid cyan line), approximate solution <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>3</mn> <mo>,</mo> <mn>12</mn> </mrow> </msub> </semantics></math> (dashed red line), and absolute errors <math display="inline"><semantics> <mo>Δ</mo> </semantics></math> of these approximations.</p>
Full article ">Figure 8
<p>Exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mi>i</mi> </msub> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mn>3</mn> </mrow> </semantics></math> (green, blue, and orange solid lines—<b>left side</b>), approximation <math display="inline"><semantics> <msub> <mi>y</mi> <mi>i</mi> </msub> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mn>3</mn> </mrow> </semantics></math> (red dashed line—<b>left side</b>), and absolute errors of these approximations (green, blue, and orange dotted lines—<b>right side</b>) for example 3.</p>
Full article ">Figure 9
<p>Exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>1</mn> </msub> </semantics></math> (solid green line), approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>1</mn> <mo>,</mo> <mn>5</mn> </mrow> </msub> </semantics></math> (dashed red line), exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>2</mn> </msub> </semantics></math> (solid blue line), approximate <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>2</mn> <mo>,</mo> <mn>5</mn> </mrow> </msub> </semantics></math> (dashed red line), exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>3</mn> </msub> </semantics></math> (solid cyan line), approximate solution <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>3</mn> <mo>,</mo> <mn>5</mn> </mrow> </msub> </semantics></math> (dashed red line), exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mn>4</mn> </msub> </semantics></math> (solid purple line), approximate solution <math display="inline"><semantics> <msub> <mi>y</mi> <mrow> <mn>4</mn> <mo>,</mo> <mn>5</mn> </mrow> </msub> </semantics></math> (dashed red line), and absolute errors <math display="inline"><semantics> <mo>Δ</mo> </semantics></math> of these approximations.</p>
Full article ">Figure 10
<p>Exact solution <math display="inline"><semantics> <msub> <mi>y</mi> <mi>i</mi> </msub> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mn>3</mn> <mo>,</mo> <mn>4</mn> </mrow> </semantics></math> (green, blue, orange, and black solid lines—<b>left side</b>), approximation <math display="inline"><semantics> <msub> <mi>y</mi> <mi>i</mi> </msub> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mn>3</mn> </mrow> </semantics></math> (red dashed line—<b>left side</b>), and absolute errors of these approximations (green, blue, orange, and black dotted lines—<b>right side</b>).</p>
Full article ">
15 pages, 301 KiB  
Article
Chosen-Ciphertext Secure Unidirectional Proxy Re-Encryption Based on Asymmetric Pairings
by Benjamin Zengin, Paulin Deupmann, Nicolas Buchmann and Marian Margraf
Appl. Sci. 2024, 14(23), 11322; https://doi.org/10.3390/app142311322 - 4 Dec 2024
Viewed by 472
Abstract
Proxy re-encryption (PRE) is a cryptographic primitive that extends public key encryption by allowing ciphertexts to be re-encrypted from one user to another without revealing information about the underlying plaintext. This makes it an essential privacy-enhancing technology, as only the intended recipient is [...] Read more.
Proxy re-encryption (PRE) is a cryptographic primitive that extends public key encryption by allowing ciphertexts to be re-encrypted from one user to another without revealing information about the underlying plaintext. This makes it an essential privacy-enhancing technology, as only the intended recipient is able to decrypt sensitive personal information. Previous PRE schemes were commonly based on symmetric bilinear pairings. However, these have been found to be slower and less secure than the more modern asymmetric pairings. To address this, we propose two new PRE scheme variants, based on the unidirectional symmetric pairing-based scheme by Weng et al. and adapted to utilize asymmetric pairings. We employ a known automated black-box reduction technique to transform the base scheme to the asymmetric setting, identify its shortcomings, and subsequently present an alternative manual transformation that fixes these flaws. The adapted schemes retain the properties of the base scheme and are therefore CCA-secure in the adaptive corruption model without the use of random oracles, while being faster, practical, and more secure overall than the base scheme. Full article
(This article belongs to the Special Issue Cryptography in Data Protection and Privacy-Enhancing Technologies)
38 pages, 1805 KiB  
Article
Functional Brain Network Disruptions in Parkinson’s Disease: Insights from Information Theory and Machine Learning
by Ömer Akgüller, Mehmet Ali Balcı and Gabriela Cioca
Diagnostics 2024, 14(23), 2728; https://doi.org/10.3390/diagnostics14232728 - 4 Dec 2024
Viewed by 502
Abstract
Objectives: This study investigates disruptions in functional brain networks in Parkinson’s Disease (PD), using advanced modeling and machine learning. Functional networks were constructed using the Nonlinear Autoregressive Distributed Lag (NARDL) model, which captures nonlinear and asymmetric dependencies between regions of interest (ROIs). Key [...] Read more.
Objectives: This study investigates disruptions in functional brain networks in Parkinson’s Disease (PD), using advanced modeling and machine learning. Functional networks were constructed using the Nonlinear Autoregressive Distributed Lag (NARDL) model, which captures nonlinear and asymmetric dependencies between regions of interest (ROIs). Key network metrics and information-theoretic measures were extracted to classify PD patients and healthy controls (HC), using deep learning models, with explainability methods employed to identify influential features. Methods: Resting-state fMRI data from the Parkinson’s Progression Markers Initiative (PPMI) dataset were used to construct NARDL-based networks. Metrics, such as Degree, Closeness, Betweenness, and Eigenvector Centrality, along with Network Entropy and Complexity, were analyzed. Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Long Short-Term Memory (LSTM) models, classified PD and HC groups. Explainability techniques, including SHAP and LIME, identified significant features driving the classifications. Results: PD patients showed reduced Closeness (22%) and Betweenness Centrality (18%). CNN achieved 91% accuracy, with Network Entropy and Eigenvector Centrality identified as key features. Increased Network Entropy indicated heightened randomness in PD brain networks. Conclusions: NARDL-based analysis with interpretable deep learning effectively distinguishes PD from HC, offering insights into neural disruptions and potential personalized treatments for PD. Full article
(This article belongs to the Special Issue Deep Learning in Medical Image Segmentation and Diagnosis)
Show Figures

Figure 1

Figure 1
<p>Outline of the methodology.</p>
Full article ">Figure 2
<p>Violin plots showing the distribution of average values for brain network metrics across Healthy Control (HC) and Parkinson’s Disease (PD) groups.</p>
Full article ">Figure 3
<p>Violin plots showing the distribution of information-theoretic measures across Healthy Control (HC) and Parkinson’s Disease (PD) groups.</p>
Full article ">Figure 4
<p>H-C Plane across Healthy Control (HC) and Parkinson’s Disease (PD) groups.</p>
Full article ">Figure 5
<p>Correlation matrix of network metrics for Healthy Control (HC) and Parkinson’s Disease (PD) groups.</p>
Full article ">Figure 6
<p>Aggregated confusion matrices with regard to 10-fold classification metric.</p>
Full article ">Figure 7
<p>SHAP summary plots of features for CNN, RNN, and LSTM models.</p>
Full article ">
40 pages, 3639 KiB  
Article
Nonparametric Testing for Information Asymmetry in the Mortgage Servicing Market
by Helmi Jedidi and Georges Dionne
Risks 2024, 12(12), 192; https://doi.org/10.3390/risks12120192 - 29 Nov 2024
Viewed by 482
Abstract
Our objective is to test for evidence of information asymmetry in the mortgage servicing market. Does the sale of mortgage servicing rights (MSR) by the initial lender to a second servicing institution unveil any residual asymmetric information? We are the first to analyze [...] Read more.
Our objective is to test for evidence of information asymmetry in the mortgage servicing market. Does the sale of mortgage servicing rights (MSR) by the initial lender to a second servicing institution unveil any residual asymmetric information? We are the first to analyze the originator’s selling choice of MSR. We use a large sample of U.S. mortgages that were securitized through the private-label channel during the period of January 2000 to December 2013 (more than 5 million observations). We propose a new nonparametric instrumental variable testing procedure to account for potential endogeneity. For robustness, we present parametric analyses to corroborate our results using instrumental variables. Our empirical results provide strong support for the presence of second-stage asymmetric information in the mortgage servicing market during the period of analysis and before the risk retention reform of 2014. Full article
(This article belongs to the Special Issue Financial Analysis, Corporate Finance and Risk Management)
Show Figures

Figure 1

Figure 1
<p>Lending and securitization processes.</p>
Full article ">Figure 2
<p>Switched decision vs. conditional probability of default.</p>
Full article ">Figure 3
<p>Instrumental-variable two-stage nonparametric estimator of switching mortgage default.</p>
Full article ">Figure 4
<p>Graphical illustration of model and empirical analysis.</p>
Full article ">Figure A1
<p>FICO scores at origination by payment type.</p>
Full article ">Figure A2
<p>FICO scores at origination by loan type.</p>
Full article ">Figure A3
<p>No/Low documentation at origination by payment type.</p>
Full article ">Figure A4
<p>Kernel density fitting of the FICO score.</p>
Full article ">Figure A5
<p>Kernel density fitting of the LTV ratio.</p>
Full article ">Figure A6
<p>Fitting of the KDE with multiple bandwidths.</p>
Full article ">Figure A7
<p>FICO score vs. conditional probability of default.</p>
Full article ">Figure A8
<p>Divorce rate vs. expected probability of mortgage default.</p>
Full article ">Figure A9
<p>Income level vs. expected probability of mortgage default.</p>
Full article ">
23 pages, 1145 KiB  
Article
State-of-the-Art Trends in Data Compression: COMPROMISE Case Study
by David Podgorelec, Damjan Strnad, Ivana Kolingerová and Borut Žalik
Entropy 2024, 26(12), 1032; https://doi.org/10.3390/e26121032 - 29 Nov 2024
Viewed by 737
Abstract
After a boom that coincided with the advent of the internet, digital cameras, digital video and audio storage and playback devices, the research on data compression has rested on its laurels for a quarter of a century. Domain-dependent lossy algorithms of the time, [...] Read more.
After a boom that coincided with the advent of the internet, digital cameras, digital video and audio storage and playback devices, the research on data compression has rested on its laurels for a quarter of a century. Domain-dependent lossy algorithms of the time, such as JPEG, AVC, MP3 and others, achieved remarkable compression ratios and encoding and decoding speeds with acceptable data quality, which has kept them in common use to this day. However, recent computing paradigms such as cloud computing, edge computing, the Internet of Things (IoT), and digital preservation have gradually posed new challenges, and, as a consequence, development trends in data compression are focusing on concepts that were not previously in the spotlight. In this article, we try to critically evaluate the most prominent of these trends and to explore their parallels, complementarities, and differences. Digital data restoration mimics the human ability to omit memorising information that is satisfactorily retrievable from the context. Feature-based data compression introduces a two-level data representation with higher-level semantic features and with residuals that correct the feature-restored (predicted) data. The integration of the advantages of individual domain-specific data compression methods into a general approach is also challenging. To the best of our knowledge, a method that addresses all these trends does not exist yet. Our methodology, COMPROMISE, has been developed exactly to make as many solutions to these challenges as possible inter-operable. It incorporates features and digital restoration. Furthermore, it is largely domain-independent (general), asymmetric, and universal. The latter refers to the ability to compress data in a common framework in a lossy, lossless, and near-lossless mode. COMPROMISE may also be considered an umbrella that links many existing domain-dependent and independent methods, supports hybrid lossless–lossy techniques, and encourages the development of new data compression algorithms. Full article
(This article belongs to the Special Issue Information Theory and Coding for Image/Video Processing)
Show Figures

Figure 1

Figure 1
<p>The COMPROMISE feature-based data compression encoding/decoding concept.</p>
Full article ">Figure 2
<p>A variant of the COMPROMISE concept with lossy compression replacing feature extraction.</p>
Full article ">Figure 3
<p>A draft of a compressive-sensing variant of the COMPROMISE concept.</p>
Full article ">Figure 4
<p>Interval features with linear approximation between two successive distinct extremes in a 1D (digital audio) example.</p>
Full article ">Figure 5
<p>An interval feature with the grid-based mask polyline approximation between two successive distinct extremes in a 1D (digital audio) example.</p>
Full article ">Figure 6
<p>Compressing digital audio with the COMPROMISE variant of replacing feature extraction with lossy compression.</p>
Full article ">
Back to TopTop