Skin cancer is a prevalent and potentially fatal disease that requires early detection for effective treatment. We trained and evaluated five YOLOv8 classification model variants (YOLOv8n-cls, YOLOv8s-cls, YOLOv8m-cls...
详细信息
ISBN:
(数字)9798350350548
ISBN:
(纸本)9798350350555
Skin cancer is a prevalent and potentially fatal disease that requires early detection for effective treatment. We trained and evaluated five YOLOv8 classification model variants (YOLOv8n-cls, YOLOv8s-cls, YOLOv8m-cls, YOLOv8l-cls, and YOLOv8x-cls) on the HAM10000 dataset, which contains 10,015 dermatoscopic images of common pigmented skin lesions. The models were trained for 30 epochs using data augmentation techniques to enhance generalization. Performance was assessed using metrics including accuracy, precision, recall, F1-score, and inference time. The YOLOv8x-cls model achieved the highest accuracy of 86.2% and precision of 82.1%, while the YOLOv81-cIs model demonstrated the best balance with the highest F1-score of 77.0%. Compared to previous ensemble approaches, our single YOLOv8 models achieved superior performance with lower computational overhead. The YOLOv8n-cls variant showed the fastest inference time of 0.5 ms, making it suitable for real-time applications. Our results demonstrate the potential of YOLOv8-based models for accurate and efficient skin lesion classification, which could aid in early skin cancer detection and improve patient outcomes.
Ensemble classifiers are constructed from various component classifiers by aggregating the results of the individual classifiers in some manner. While there are many ways to do this, one of the least computationally i...
Ensemble classifiers are constructed from various component classifiers by aggregating the results of the individual classifiers in some manner. While there are many ways to do this, one of the least computationally intensive is to use a voting ensemble, where the vote of each classifier is weighted by a number representing its performance. Choosing the appropriate weights is an involved process, that ranges from using trial and error to using sophisticated machine learning algorithms. In this paper, we proposed a novel technique that uses generative adversarial networks to customize the weights used for each individual input to the ensemble. Specifically, we propose to use conditional generative adversarial networks to generate the weights, with the input image acting as the label to the network and the appropriate weight being the output. We test the proposed system on an image classification task and show that the proposed method produces better results than using fixed weights by 2.43 percent.
This paper describes modified robust algorithms for a line clipping by a convex polygon in E2 and a convex polyhedron in E3. The proposed algorithm is based on the Cyrus-Beck algorithm and uses homogeneous coordinates...
详细信息
The research paper highlights the remarkable threats that plant diseases pose to global agricultural productivity and food security. The early detection and accurate recognition of these diseases are crucial for the e...
详细信息
The use of diagram-based development approach especially Model Driven Development (MDD), where the executable code can be generated directly from the diagrammatic system model, has gained much popularity in the recent...
The use of diagram-based development approach especially Model Driven Development (MDD), where the executable code can be generated directly from the diagrammatic system model, has gained much popularity in the recent years. The use of automated code generation allows to focus primarily on the core design aspects of system design without having to deal with the developmental challenges. Moreover, the approach also avoids any gap between the system design and the developed system, which can be very common when code development is manual. However, at present, these generated programs have to be manually checked for semantics correctness and hence, the possibility of an automated model level verification of such MDD artifacts is the next big step in this domain. There are multiple tools today that implement MDD and are being used by the research world for exploring the possible verification alternatives. In this paper we look closely into the verification of one such standardized profile, UML for Real-Time (UML-RT), developed particularly for MDD of Real-Time systems. Eclipse Papyrus-RT is an open-source modeling tool that supports UML-RT for specifying software systems. nuXmv is a model checker, that can verify system models against a set of properties defined using temporal logics. This paper presents a system that automatically translates Papyrus-RT models into a behaviorally equivalent nuXmv model.
Constructing portfolio by proper asset selection to maximize return and minimize risk has been considered an essential task for investment activities. Rich portfolio optimizations with realistic constraints are NP-har...
详细信息
Heuristic algorithms are dependent on many coefficients like the number of iterations or individuals. However, quite often these algorithms move individuals toward the best in the population. Based on this observation...
Heuristic algorithms are dependent on many coefficients like the number of iterations or individuals. However, quite often these algorithms move individuals toward the best in the population. Based on this observation, we propose the idea of federated heuristics. The proposed idea is to initially distribute individuals into certain intervals. Then, it performs a specified number of iterations of the algorithm to identify the potentially best intervals. Sorted intervals (in relation to the best-adapted individual) make it possible to separate the appropriate size of the population in each of them. Moreover, these clusters are merged by a fuzzy algorithm due to a decrease in their numbers. The more significant the interval, the greater the number of individuals and iterations allocated in these areas. As a consequence, several instances of the selected heuristic algorithm are triggered, which can divide the best individual. The proposed technique was described using the red fox algorithm and tested at a classic set of functions with different parameters of the used heuristic.
Spaced repetition is a learning task that helps users repeat learning items at optimally computed intervals. It is best suited for small learning units such as learning cards. In a typical learning scenario these lear...
Spaced repetition is a learning task that helps users repeat learning items at optimally computed intervals. It is best suited for small learning units such as learning cards. In a typical learning scenario these learning cards do not stand by themselves, rather they are the last step in the students’ work process, summarizing the students’ interaction with the learning material. In this paper a framework is developed that allows learning material curators and students to create spaced repetition learning units with back-references to multimedia learning content, i.e., learning videos. With the advent of video learning nuggets, a wealth of content was created that can now be used in spaced repetition learning. The approach also allows for linking to specific parts of those learning videos to enable students to quickly review learning content specific to the learning unit currently worked on.
In this paper we present the mathematical formulation of the problem of planning the Asset Liability Management if the Presence of Normative and Internal Constraints. As a nonlinear optimal control problem, it has spe...
详细信息
A large number of objects participating in the voting can be an advantage as well as a disadvantage. In the case of decentralized federated learning, adding the model to the aggregation is preceded by a vote. The choi...
A large number of objects participating in the voting can be an advantage as well as a disadvantage. In the case of decentralized federated learning, adding the model to the aggregation is preceded by a vote. The choice of voters and their results can be falsified through various attacks such as dataset poisoning. In this paper, we propose a fuzzy consensus analyzing the results of individual voters regarding the aggregation of a given model. The consensus is based on a fuzzy controller that selects the most reliable models for aggregation. For this reason, it uses image-modifying heuristics and quick evaluations of incoming results. If a decision is made that a selected client is unreliable several times, it is blocked to reduce the number of performed operations. The proposed system was tested on selected tasks related to image classification. The results were discussed and compared to evaluate the proposed system.
暂无评论