Informaatioteknologian tiedekuntahttps://jyx.jyu.fi/handle/123456789/255932024-03-29T14:03:22Z2024-03-29T14:03:22ZCreative and Adversarial Cellular Automata for Simulating Resilience in Industry 5.0Terziyan, VaganTerziian, ArturVitko, Oleksandrahttps://jyx.jyu.fi/handle/123456789/941132024-03-28T14:17:33Z2024-03-28T10:19:20ZCreative and Adversarial Cellular Automata for Simulating Resilience in Industry 5.0
Terziyan, Vagan; Terziian, Artur; Vitko, Oleksandra
Emerging Industry 5.0 pushes advanced automation towards resilient solutions with enhanced human role. Resilience as an ability to sustain processes in the face of disruptions and adversarial attacks requires careful modelling and simulation. Cellular automata are efficient mathematical models used to simulate the behavior of complex systems, which change their state based on a set of predefined rules. In this paper, we suggest several updates to cellular automata (particularly Conway's “Game of Life”) to address resilience. These include “Life and Creation”, “War and Peace”, and their hybrid “War and Creation” capable of addressing the important components of resilience, such as controllable creativity and adversarial interactions. Inherited in these updates and known advantages of cellular automata, such as simplicity, emergent behavior, parallelism, and adaptability, makes it a powerful simulation tool for a wide range of Industry 5.0 systems that involve humans, smart infrastructure, their complex and adversarial interactions, safety, and resilience.
See presentation slides: https://ai.it.jyu.fi/ISM-2023-Cellular_Automata.pptx
2024-03-28T10:19:20ZDeep Neural Networks, Cellular Automata and Petri Nets : Useful Hybrids for Smart ManufacturingKaikova, OlenaTerziyan, Vaganhttps://jyx.jyu.fi/handle/123456789/940992024-03-27T14:17:23Z2024-03-27T13:03:51ZDeep Neural Networks, Cellular Automata and Petri Nets : Useful Hybrids for Smart Manufacturing
Kaikova, Olena; Terziyan, Vagan
In the era of Industry 4.0 and beyond, intelligent and reliable models are vital for processes and assets. Models in smart manufacturing involve combining knowledge-based and data-driven methods with discrete and continuous modelling components. Formalism choice determines models' strengths and weaknesses in accuracy, efficiency, robustness, and explainability. Hybrid models seem to be the only way to address the complexity of modern industrial systems with respect to different and conflicting quality criteria. This study focuses on three paradigms: Petri nets, cellular automata, and neural network driven deep learning. We create four hybrids: Petri nets controlling deep neural networks, and vice versa; cellular automata controlling deep neural networks, and vice versa. These hybrids combine explainable discrete models with continuous black-box models, enhancing either explainability with robustness or elevating accuracy with efficiency. The flexibility of these and similar hybrids enable enhancement of the scope and quality of modeling and simulation in smart manufacturing.
See presentation slides: https://ai.it.jyu.fi/ISM-2023-Hybrids.pptx
2024-03-27T13:03:51ZTaxonomy-Informed Neural Networks for Smart ManufacturingTerziyan, VaganVitko, Oleksandrahttps://jyx.jyu.fi/handle/123456789/940972024-03-27T14:16:36Z2024-03-27T12:44:21ZTaxonomy-Informed Neural Networks for Smart Manufacturing
Terziyan, Vagan; Vitko, Oleksandra
A neural network (NN) is known to be an efficient and learnable tool supporting decision-making processes particularly in Industry 4.0. The majority of NNs are data-driven and, therefore, depend on training data quantity and quality. The current trend in enhancing data-driven models with knowledge-based models promises to enable effective NNs with less data. So-called physics-informed NNs use additional knowledge from computational science to improve NN training. Quite much of the knowledge is available as logical constraints from domain ontologies, and NNs may benefit from using it. In this paper, we study the concept of Taxonomy-Informed NN (TINN), which combines data-driven training of NNs with ontological knowledge. We study different patterns of NN training with additional knowledge on class-subclass hierarchies and instance-class relationships with potential for federated learning. Our experiments show that additional knowledge, which influences TINNs’ training process through the loss function at backpropagation, improves the quality of trained models.
See presentation slides: https://ai.it.jyu.fi/ISM-2023-TINN.pptx
2024-03-27T12:44:21ZUsing Cloning-GAN Architecture to Unlock the Secrets of Smart Manufacturing : Replication of Cognitive ModelsTerziyan, VaganTiihonen, Timohttps://jyx.jyu.fi/handle/123456789/940962024-03-27T14:16:05Z2024-03-27T12:38:50ZUsing Cloning-GAN Architecture to Unlock the Secrets of Smart Manufacturing : Replication of Cognitive Models
Terziyan, Vagan; Tiihonen, Timo
As Industry 4.0 and 5.0 evolve to be highly automated but human-centric, there is a need for process modeling based on digital replicas of physical objects including humans. Knowledge distillation and cognitive cloning offer a way to train operational copies of decision-making black boxes, or donors, without requiring additional data. In this paper, we propose an architecture and analytics for a generative adversarial network, called Cloning-GAN, which enables donor-clone knowledge transfer, including the donor's individual biases. The architecture involves generating challenging samples to be labeled by the donor and used as training data for the clone. We consider several multicriteria requirements for the generated data, including closeness to the decision boundary, uniform distribution in the decision space, maximal confusion for the donor, and challenge for the clone. We present various strategies to balance these conflicting criteria forcing the clone learning quickly the hidden cognitive skills and biases of the donor.
See presentation slides: https://ai.it.jyu.fi/ISM-2023-Cloning_GAN.pptx
2024-03-27T12:38:50ZHybrid modeling design patternsRudolph, MajaKurz, StefanRakitsch, Barbarahttps://jyx.jyu.fi/handle/123456789/940952024-03-27T14:19:04Z2024-03-27T12:35:26ZHybrid modeling design patterns
Rudolph, Maja; Kurz, Stefan; Rakitsch, Barbara
Design patterns provide a systematic way to convey solutions to recurring modeling challenges. This paper introduces design patterns for hybrid modeling, an approach that combines modeling based on first principles with data-driven modeling techniques. While both approaches have complementary advantages there are often multiple ways to combine them into a hybrid model, and the appropriate solution will depend on the problem at hand. In this paper, we provide four base patterns that can serve as blueprints for combining data-driven components with domain knowledge into a hybrid approach. In addition, we also present two composition patterns that govern the combination of the base patterns into more complex hybrid models. Each design pattern is illustrated by typical use cases from application areas such as climate modeling, engineering, and physics.
2024-03-27T12:35:26ZShowrooming Behavior, Omnichannel Self-Efficacy, and Perceived Channel Integration as Antecedents of Revisit IntentionHolkkola, MatildaNyrhinen, JussiMakkonen, MarkusFrank, Laurihttps://jyx.jyu.fi/handle/123456789/940832024-03-26T14:15:57Z2024-03-26T08:33:33ZShowrooming Behavior, Omnichannel Self-Efficacy, and Perceived Channel Integration as Antecedents of Revisit Intention
Holkkola, Matilda; Nyrhinen, Jussi; Makkonen, Markus; Frank, Lauri
This study investigates how consumers’ omnichannel self-efficacy and showrooming behavior affect the perceived channel integration of a retailer and how perceived channel integration affects consumers’ revisit intention. In this study, showrooming behavior includes consumers first engaging with products in brick-and-mortar (B&M) stores and then searching for additional information for potential purchases online on the same or a competitive retailer’s online channels. Because competitive showrooming is common, B&M retailers have an interest in integrating their channels to offer a seamless shopping experience for showroomers to attract and retain possible customers. We hypothesize that omnichannel self-efficacy positively influences consumers’ showrooming behavior and the perceived channel integration of offline and online channels. We also hypothesize that showrooming behavior positively affects perceived channel integration and, ultimately, perceived channel integration positively affects consumers’ revisit intention. The survey data consists of 1,028 Finnish omnichannel consumers. We used partial least squares structural equation modeling to test our hypotheses, which were all supported. As a novel finding, omnichannel self-efficacy and showrooming behavior are found as antecedents of perceived channel integration. The practical implications are that B&M retailers with an omnichannel-skilled customer base should link their online channels in their B&M stores to reduce competitive showrooming.
2024-03-26T08:33:33ZPredicting pathological complete response based on weakly and semi-supervised joint learning in breast cancer multi-parametric MRIHao, XinyuXu, HongmingZhao, NannanYu, TaoHämäläinen, TimoCong, Fengyuhttps://jyx.jyu.fi/handle/123456789/940012024-03-21T14:15:49Z2024-03-21T07:32:25ZPredicting pathological complete response based on weakly and semi-supervised joint learning in breast cancer multi-parametric MRI
Hao, Xinyu; Xu, Hongming; Zhao, Nannan; Yu, Tao; Hämäläinen, Timo; Cong, Fengyu
Neoadjuvant chemotherapy (NAC) is the primary treatment used to reduce the tumor size in early breast cancer. Patients who achieve a pathological complete response (pCR) after NAC treatment have a significantly higher five-year survival rate. However, accurately predicting whether patients could achieve pCR remains challenging due to the limited availability of manually annotated MRI data. This study develops a weakly and semi-supervised joint learning model that integrates multi-parametric MR images to predict pCR to NAC in breast cancer patients. First, the attention-based multi-instance learning model is designed to characterize the representation of multi-parametric MR images in a weakly supervised learning setting. The Mean-Teacher learning framework is then developed to locate tumor regions for extracting radiochemical parameters in a semi-supervised learning setting. Finally, all extracted MR imaging features are fused to predict pCR to NAC. Our experiments were conducted on a cohort of 442 patients with multi-parametric MR images and NAC outcomes. The results demonstrate that our proposed model, which leverages multi-parametric MRI data, provides the AUC value of over 0.85 in predicting pCR to NAC, outperforming other comparative methods.
2024-03-21T07:32:25ZSimultaneously segmenting and classifying cell nuclei by using multi-task learning in multiplex immunohistochemical tissue microarray sectionsWang, RanranQiu, YusongHao, XinyuJin, ShanGao, JunxiuQi, HengXu, QiZhang, YongXu, Hongminghttps://jyx.jyu.fi/handle/123456789/940002024-03-21T14:16:44Z2024-03-21T07:16:56ZSimultaneously segmenting and classifying cell nuclei by using multi-task learning in multiplex immunohistochemical tissue microarray sections
Wang, Ranran; Qiu, Yusong; Hao, Xinyu; Jin, Shan; Gao, Junxiu; Qi, Heng; Xu, Qi; Zhang, Yong; Xu, Hongming
Quantitative analysis of tumor immune microenvironment (TIME) in immunohistochemical (IHC) tissue microarray (TMA) sections is crucial in diagnosis and treatment recommendations for cancer patients. Nuclei segmentation and classification are the prerequisites for the TIME quantification, but it still lacks of robust nuclear quantification models used for IHC histological slides. In this paper, we design an approach for simultaneously segmenting and classifying cell nuclei in multiplex IHC TMA sections. The large TMA tissue core is first divided into a set of small overlapping patches, where cell nuclei are then simultaneously segmented and classified by using our multi-task learning model. The model has one feature encoder with cascaded separable-ResUnit blocks, and three decoder branches that incorporate the Self-Attention modules and DenseUnit blocks to perform nuclear segmentation, classification and distance map regression, respectively. After processing all patches, the weighted loss map and vote mechanism are applied to seamlessly stitch patch-level predictions to form the tissue core level results. We finally exploit generalized Laplacian of Gaussian (gLoG) filters based algorithm to post-process segmentation results to further split overlapping cell nuclei. Quantitative evaluations have been performed on a IHC stained histological image dataset with 9725 manually identified cell nuclei and a public H&E stained dataset (CoNSep), which show that our model outperforms state-of-the-art nuclei segmentation and classification models. The qualitative evaluations on TMA sections show the potential of using our approach in clinical applications.
2024-03-21T07:16:56ZSystem-Information Models of Digital TwinsKorablyov, MykolaLutskyy, SergeyIvanisenko, IhorFomichov, Oleksandrhttps://jyx.jyu.fi/handle/123456789/939792024-03-20T14:20:08Z2024-03-20T08:53:53ZSystem-Information Models of Digital Twins
Korablyov, Mykola; Lutskyy, Sergey; Ivanisenko, Ihor; Fomichov, Oleksandr
Nagar, Atulya K.; Singh Jat, Dharm; Mishra, Durgesh; Joshi, Amit
To represent the production process, a digital twin model is used, which takes into account the real parameters of technological processes. Management of product life cycle processes is implemented on the basis of a digital twin of the Unified System Information Space (USIS), built on system-information models of processes and systems. It is used as a platform for management using software products Product Lifecycle System Information (PLSI), which are system-compatible with technological system software Product Lifecycle Management (PLM). The digital twin describes the functional dependence of the expanded uncertainty of the normalized information space on the values of the nominal parameters for a specific production technology using a software product (USIS + PLSI + PLM). This allows you to use software products for designing CAD, CAM, and CAE systems when solving production problems on one information platform. Using a system-information approach to modeling digital twins of production allows you to effectively solve problems related to the analysis, synthesis, management, and forecasting of production.
2024-03-20T08:53:53ZEffect of variable selection strategy on the predictive models for adverse pregnancy outcomes of pre-eclampsia : A retrospective studyZheng, DongyingHao, XinyuKhan, MuhanmmadKang, FuliLi, FanHämäläinen, TimoWang, Lixiahttps://jyx.jyu.fi/handle/123456789/939692024-03-19T14:19:38Z2024-03-19T12:28:30ZEffect of variable selection strategy on the predictive models for adverse pregnancy outcomes of pre-eclampsia : A retrospective study
Zheng, Dongying; Hao, Xinyu; Khan, Muhanmmad; Kang, Fuli; Li, Fan; Hämäläinen, Timo; Wang, Lixia
Objectives: The improvement of prediction for adverse pregnancy outcomes is quite essential to the women suffering from pre-eclampsia, while the collection of predictive indicators is the prerequisite. The traditional knowledge-based strategy for variable selection confronts challenge referring to dataset with high-dimensional or unfamiliar data. In this study, we employed five different automatic variable selection methods to screen out influential indicators, and evaluated the performance of constructed predictive models. Methods: Seven hundreds and thirty-three Han-Chinese women were enrolled and 56 clinical and laboratory variables were recorded. After grouping based on binary pregnancy outcomes, statistical description and analysis were performed. Then, utilizing forward stepwise logistic regression (FSLR) as the reference method, another four variable selection strategies were included for filtering contributing variables as the predictive subsets, respectively. Finally, the logistic regression prediction models were constructed by the five subsets and evaluated by the receiver operator characteristic curve. Results: The variables confirmed statistical significance between the adverse and satisfactory outcomes groups did not overlap with the variables selected by selection strategies. “Platelet” and “Creatinine clearance rate” were the most influential indicator to predict adverse maternal outcome, while “Birth weight of neonates” was the best indicator for predicting adverse neonatal outcome. In average, the predictive models for neonatal outcomes achieved better performance than models for maternal outcomes. “Mutual information” and “Recursive feature elimination” were the best strategy under current dataset and study design. Conclusions: Variable selection strategies may provide an alternative approach besides picking influential indicators by statistical significance. Future work will focus on applying different variable selection methods to the high-dimensional dataset, which includes novel or unfamiliar variables. This aims to identify the most appropriate collection of predictors that can enhance prediction ability and clinical decision-making.
2024-03-19T12:28:30Z