计算机科学最新文献

英文中文

IF： -

Post-earthquake structural damage detection with tunable semi-synthetic image generation

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2025-02-22 DOI: 10.1016/j.engappai.2025.110302

Piercarlo Dondi , Alessio Gullotti , Michele Inchingolo , Ilaria Senaldi , Chiara Casarotti , Luca Lombardi , Marco Piastra

In the aftermath of an earthquake, conducting rapid structural safety assessments is essential. A Deep Learning-based damage detector capable of automatically analyzing videos from Unmanned Aircraft Systems (UAS) surveys would be highly beneficial for this purpose. Despite significant advancements in object detection using Deep Convolutional Neural Networks (DCNNs), developing an effective post-earthquake damage detector remains challenging due to the scarcity of large, annotated image datasets. In this work, we present a system to create a large number of images where artificial damage instances are applied to real-world three-dimensional (3D) models of buildings and bridges. We defined such images as semi-synthetic. The proposed method relies on the definition, made by human experts, of meta-annotations from which a variety of damage instances can be generated in a controlled way. Semi-synthetic images are designed to augment real-world datasets, enhancing the training process of a DCNN-based damage detector. This semi-synthetic image augmentation can be iteratively refined to target the most critical cases. Experiments conducted on the ‘Image Database for Earthquake damage Annotation’ (IDEA) dataset shown that a detector trained on a combination of real and semi-synthetic images performs better than one trained on real images alone. A damage detector trained using the proposed strategy was then incorporated into a system that analyzes and tracks multiple damage instances in UAS-acquired videos, generating concise summaries of the findings. The effectiveness of the system was validated by the analysis of post-earthquake UAS videos and the production of reports that were reviewed by structural engineering experts.

{"title":"Post-earthquake structural damage detection with tunable semi-synthetic image generation","authors":"Piercarlo Dondi , Alessio Gullotti , Michele Inchingolo , Ilaria Senaldi , Chiara Casarotti , Luca Lombardi , Marco Piastra","doi":"10.1016/j.engappai.2025.110302","DOIUrl":"10.1016/j.engappai.2025.110302","url":null,"abstract":"<div><div>In the aftermath of an earthquake, conducting rapid structural safety assessments is essential. A Deep Learning-based damage detector capable of automatically analyzing videos from Unmanned Aircraft Systems (UAS) surveys would be highly beneficial for this purpose. Despite significant advancements in object detection using Deep Convolutional Neural Networks (DCNNs), developing an effective post-earthquake damage detector remains challenging due to the scarcity of large, annotated image datasets. In this work, we present a system to create a large number of images where artificial damage instances are applied to real-world three-dimensional (3D) models of buildings and bridges. We defined such images as semi-synthetic. The proposed method relies on the definition, made by human experts, of meta-annotations from which a variety of damage instances can be generated in a controlled way. Semi-synthetic images are designed to augment real-world datasets, enhancing the training process of a DCNN-based damage detector. This semi-synthetic image augmentation can be iteratively refined to target the most critical cases. Experiments conducted on the ‘Image Database for Earthquake damage Annotation’ (IDEA) dataset shown that a detector trained on a combination of real and semi-synthetic images performs better than one trained on real images alone. A damage detector trained using the proposed strategy was then incorporated into a system that analyzes and tracks multiple damage instances in UAS-acquired videos, generating concise summaries of the findings. The effectiveness of the system was validated by the analysis of post-earthquake UAS videos and the production of reports that were reviewed by structural engineering experts.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"147 ","pages":"Article 110302"},"PeriodicalIF":7.5,"publicationDate":"2025-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143471370","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Goals as reward-producing programs

IF 23.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Nature Machine Intelligence

Pub Date : 2025-02-21 DOI: 10.1038/s42256-025-00981-4

Guy Davidson, Graham Todd, Julian Togelius, Todd M. Gureckis, Brenden M. Lake

People are remarkably capable of generating their own goals, beginning with child’s play and continuing into adulthood. Despite considerable empirical and computational work on goals and goal-oriented behaviour, models are still far from capturing the richness of everyday human goals. Here we bridge this gap by collecting a dataset of human-generated playful goals (in the form of scorable, single-player games), modelling them as reward-producing programs and generating novel human-like goals through program synthesis. Reward-producing programs capture the rich semantics of goals through symbolic operations that compose, add temporal constraints and allow program execution on behavioural traces to evaluate progress. To build a generative model of goals, we learn a fitness function over the infinite set of possible goal programs and sample novel goals with a quality-diversity algorithm. Human evaluators found that model-generated goals, when sampled from partitions of program space occupied by human examples, were indistinguishable from human-created games. We also discovered that our model’s internal fitness scores predict games that are evaluated as more fun to play and more human-like.

{"title":"Goals as reward-producing programs","authors":"Guy Davidson, Graham Todd, Julian Togelius, Todd M. Gureckis, Brenden M. Lake","doi":"10.1038/s42256-025-00981-4","DOIUrl":"https://doi.org/10.1038/s42256-025-00981-4","url":null,"abstract":"<p>People are remarkably capable of generating their own goals, beginning with child’s play and continuing into adulthood. Despite considerable empirical and computational work on goals and goal-oriented behaviour, models are still far from capturing the richness of everyday human goals. Here we bridge this gap by collecting a dataset of human-generated playful goals (in the form of scorable, single-player games), modelling them as reward-producing programs and generating novel human-like goals through program synthesis. Reward-producing programs capture the rich semantics of goals through symbolic operations that compose, add temporal constraints and allow program execution on behavioural traces to evaluate progress. To build a generative model of goals, we learn a fitness function over the infinite set of possible goal programs and sample novel goals with a quality-diversity algorithm. Human evaluators found that model-generated goals, when sampled from partitions of program space occupied by human examples, were indistinguishable from human-created games. We also discovered that our model’s internal fitness scores predict games that are evaluated as more fun to play and more human-like.</p>","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"15 1","pages":""},"PeriodicalIF":23.8,"publicationDate":"2025-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143462924","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

NNBSVR: Neural Network-Based Semantic Vector Representations of ICD-10 codes

IF 3.4 2区计算机科学 Q2 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Intelligence

Pub Date : 2025-02-21 DOI: 10.1007/s10489-025-06349-w

Monah Bou Hatoum, Jean Claude Charr, Alia Ghaddar, Christophe Guyeux, David Laiymani

Automatically predicting ICD-10 codes from clinical notes using machine learning models can reduce the burden of manual coding. However, existing methods often overlook the semantic relationships between ICD-10 codes, resulting in inaccurate evaluations when clinically similar codes are considered completely different. Traditional evaluation metrics, which rely on equality-based matching, fail to capture the clinical relevance of predicted codes. This study introduces NNBSVR (Neural Network-Based Semantic Vector Representations), a novel approach for generating semantic-based vector representations of ICD-10 codes. Unlike traditional approaches that rely on exact code matching, NNBSVR incorporates contextual and hierarchical information to enhance both prediction accuracy and evaluation methods. We validate NNBSVR using intrinsic and extrinsic evaluation methods. Intrinsic evaluation assesses the vectors’ ability to reconstruct the ICD-10 hierarchy and identify clinically meaningful clusters. Extrinsic evaluation compares our relevancy-based approach, which includes customized evaluation metrics, to traditional equality-based metrics on an ICD-10 code prediction task using a 9.57 million clinical notes corpus. NNBSVR demonstrates significant improvements over equality-based metrics, achieving a 9.81% gain in micro-F1 score on the training set and a 12.73% gain on the test set. A manual review by medical experts on a sample of 10,000 predictions confirms an accuracy of 92.58%, further validating our approach. This study makes two significant contributions: first, the development of semantic-based vector representations that encapsulate ICD-10 code relationships and context; second, the customization of evaluation metrics to incorporate clinical relevance. By addressing the limitations of traditional equality-based evaluation metrics, NNBSVR enhances the automated assignment of ICD-10 codes in clinical settings, demonstrating superior performance over existing methods.

{"title":"NNBSVR: Neural Network-Based Semantic Vector Representations of ICD-10 codes","authors":"Monah Bou Hatoum, Jean Claude Charr, Alia Ghaddar, Christophe Guyeux, David Laiymani","doi":"10.1007/s10489-025-06349-w","DOIUrl":"10.1007/s10489-025-06349-w","url":null,"abstract":"<div><p>Automatically predicting ICD-10 codes from clinical notes using machine learning models can reduce the burden of manual coding. However, existing methods often overlook the semantic relationships between ICD-10 codes, resulting in inaccurate evaluations when clinically similar codes are considered completely different. Traditional evaluation metrics, which rely on equality-based matching, fail to capture the clinical relevance of predicted codes. This study introduces <i>NNBSVR</i> (Neural Network-Based Semantic Vector Representations), a novel approach for generating semantic-based vector representations of ICD-10 codes. Unlike traditional approaches that rely on exact code matching, <i>NNBSVR</i> incorporates contextual and hierarchical information to enhance both prediction accuracy and evaluation methods. We validate <i>NNBSVR</i> using intrinsic and extrinsic evaluation methods. Intrinsic evaluation assesses the vectors’ ability to reconstruct the ICD-10 hierarchy and identify clinically meaningful clusters. Extrinsic evaluation compares our relevancy-based approach, which includes customized evaluation metrics, to traditional equality-based metrics on an ICD-10 code prediction task using a 9.57 million clinical notes corpus. <i>NNBSVR</i> demonstrates significant improvements over equality-based metrics, achieving a 9.81% gain in micro-F1 score on the training set and a 12.73% gain on the test set. A manual review by medical experts on a sample of 10,000 predictions confirms an accuracy of 92.58%, further validating our approach. This study makes two significant contributions: first, the development of semantic-based vector representations that encapsulate ICD-10 code relationships and context; second, the customization of evaluation metrics to incorporate clinical relevance. By addressing the limitations of traditional equality-based evaluation metrics, <i>NNBSVR</i> enhances the automated assignment of ICD-10 codes in clinical settings, demonstrating superior performance over existing methods.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 6","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143455535","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Optimal Control Using IsoCost-Based Dynamic Programming

IF 2.2 4区计算机科学 Q2 AUTOMATION & CONTROL SYSTEMS

IET Control Theory and Applications

Pub Date : 2025-02-21 DOI: 10.1049/cth2.70014

Fatemeh Alvankarian, Ahmad Kalhor, Mehdi Tale Masouleh

In this paper, a novel data-driven optimal control method based on reinforcement learning concepts is introduced. The proposed algorithm performs as a workaround to solving the Hamilton–Jacobi–Bellman equation. The main concept behind the proposed algorithm is the so-called IsoCost hypersurface (ICHS), which is a hypersurface in the state space of the system formed by points from which a specific amount of cost is spent by the control strategy in order to asymptotically stabilize the system. The fact that the control strategy requires to spend equal costs in order to stabilize all points on an ICHS is the reason for the naming of the IsoCost concept. Additional assumptions and definitions are mentioned before providing the theory of ICHS optimality. This theory proves, by contradiction, that the ICHS corresponding to the optimal control policy surrounds the ICHSs corresponding to other non-optimal control solutions. This paves the path to finding the optimal control solution using dynamic programming. The proposed method is implemented on the linear, fixed-base inverted pendulum, cart-pole and torsional pendulum bar system models and the results are compared with that of literature. The performance of this method in terms of cost, settling time and computation time is shown using numeric and illustrative comparisons.

{"title":"Optimal Control Using IsoCost-Based Dynamic Programming","authors":"Fatemeh Alvankarian, Ahmad Kalhor, Mehdi Tale Masouleh","doi":"10.1049/cth2.70014","DOIUrl":"https://doi.org/10.1049/cth2.70014","url":null,"abstract":"<p>In this paper, a novel data-driven optimal control method based on reinforcement learning concepts is introduced. The proposed algorithm performs as a workaround to solving the Hamilton–Jacobi–Bellman equation. The main concept behind the proposed algorithm is the so-called IsoCost hypersurface (ICHS), which is a hypersurface in the state space of the system formed by points from which a specific amount of cost is spent by the control strategy in order to asymptotically stabilize the system. The fact that the control strategy requires to spend equal costs in order to stabilize all points on an ICHS is the reason for the naming of the IsoCost concept. Additional assumptions and definitions are mentioned before providing the theory of ICHS optimality. This theory proves, by contradiction, that the ICHS corresponding to the optimal control policy surrounds the ICHSs corresponding to other non-optimal control solutions. This paves the path to finding the optimal control solution using dynamic programming. The proposed method is implemented on the linear, fixed-base inverted pendulum, cart-pole and torsional pendulum bar system models and the results are compared with that of literature. The performance of this method in terms of cost, settling time and computation time is shown using numeric and illustrative comparisons.</p>","PeriodicalId":50382,"journal":{"name":"IET Control Theory and Applications","volume":"19 1","pages":""},"PeriodicalIF":2.2,"publicationDate":"2025-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1049/cth2.70014","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143456139","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Series Editorial: Wireless And Radio Communications

IF 11.2 2区计算机科学 Q1 ENGINEERING, ELECTRICAL & ELECTRONIC

IEEE Communications Magazine

Pub Date : 2025-02-21 DOI: 10.1109/mcom.2025.10897940

Todor Cooklev, Leif Wilhelmsson, Peiying Zhu

引用次数: 0

CNN-aided Self-Interference Estimation for In-Band Full-Duplex Systems

IF 8.6 1区计算机科学 Q1 TELECOMMUNICATIONS

IEEE Transactions on Cognitive Communications and Networking

Pub Date : 2025-02-21 DOI: 10.1109/tccn.2025.3544838

Iñigo Bilbao, Eneko Iradier, Jon Montalban, Pablo Angueira, Zhihong Hong, Yiyan Wu