Siga este enlace para ver otros tipos de publicaciones sobre el tema: Shortcut learning.
Artículos de revistas sobre el tema "Shortcut learning"
Crea una cita precisa en los estilos APA, MLA, Chicago, Harvard y otros
Consulte los 50 mejores artículos de revistas para su investigación sobre el tema "Shortcut learning".
Junto a cada fuente en la lista de referencias hay un botón "Agregar a la bibliografía". Pulsa este botón, y generaremos automáticamente la referencia bibliográfica para la obra elegida en el estilo de cita que necesites: APA, MLA, Harvard, Vancouver, Chicago, etc.
También puede descargar el texto completo de la publicación académica en formato pdf y leer en línea su resumen siempre que esté disponible en los metadatos.
Explore artículos de revistas sobre una amplia variedad de disciplinas y organice su bibliografía correctamente.
1
Kim, Doyoung, Dongmin Park, Yooju Shin, Jihwan Bang, Hwanjun Song, and Jae-Gil Lee. "Adaptive Shortcut Debiasing for Online Continual Learning." Proceedings of the AAAI Conference on Artificial Intelligence 38, no. 12 (2024): 13122–31. http://dx.doi.org/10.1609/aaai.v38i12.29211.
Texto completoResumen
We propose a novel framework DropTop that suppresses the shortcut bias in online continual learning (OCL) while being adaptive to the varying degree of the shortcut bias incurred by continuously changing environment. By the observed high-attention property of the shortcut bias, highly-activated features are considered candidates for debiasing. More importantly, resolving the limitation of the online environment where prior knowledge and auxiliary data are not ready, two novel techniques---feature map fusion and adaptive intensity shifting---enable us to automatically determine the appropriate level and proportion of the candidate shortcut features to be dropped. Extensive experiments on five benchmark datasets demonstrate that, when combined with various OCL algorithms, DropTop increases the average accuracy by up to 10.4% and decreases the forgetting by up to 63.2%.
2
Nauta, Meike, Ricky Walsh, Adam Dubowski, and Christin Seifert. "Uncovering and Correcting Shortcut Learning in Machine Learning Models for Skin Cancer Diagnosis." Diagnostics 12, no. 1 (2021): 40. http://dx.doi.org/10.3390/diagnostics12010040.
Texto completoResumen
Machine learning models have been successfully applied for analysis of skin images. However, due to the black box nature of such deep learning models, it is difficult to understand their underlying reasoning. This prevents a human from validating whether the model is right for the right reasons. Spurious correlations and other biases in data can cause a model to base its predictions on such artefacts rather than on the true relevant information. These learned shortcuts can in turn cause incorrect performance estimates and can result in unexpected outcomes when the model is applied in clinical practice. This study presents a method to detect and quantify this shortcut learning in trained classifiers for skin cancer diagnosis, since it is known that dermoscopy images can contain artefacts. Specifically, we train a standard VGG16-based skin cancer classifier on the public ISIC dataset, for which colour calibration charts (elliptical, coloured patches) occur only in benign images and not in malignant ones. Our methodology artificially inserts those patches and uses inpainting to automatically remove patches from images to assess the changes in predictions. We find that our standard classifier partly bases its predictions of benign images on the presence of such a coloured patch. More importantly, by artificially inserting coloured patches into malignant images, we show that shortcut learning results in a significant increase in misdiagnoses, making the classifier unreliable when used in clinical practice. With our results, we, therefore, want to increase awareness of the risks of using black box machine learning models trained on potentially biased datasets. Finally, we present a model-agnostic method to neutralise shortcut learning by removing the bias in the training dataset by exchanging coloured patches with benign skin tissue using image inpainting and re-training the classifier on this de-biased dataset.
3
Geirhos, Robert, Jörn-Henrik Jacobsen, Claudio Michaelis, et al. "Shortcut learning in deep neural networks." Nature Machine Intelligence 2, no. 11 (2020): 665–73. http://dx.doi.org/10.1038/s42256-020-00257-z.
Texto completo
4
Fay, Louisa, Erick Cobos, Bin Yang, Sergios Gatidis, and Thomas Küstner. "Avoiding Shortcut-Learning by Mutual Information Minimization in Deep Learning-Based Image Processing." IEEE Access 11 (2023): 64070–86. http://dx.doi.org/10.1109/access.2023.3289397.
Texto completo
5
POTAPOV, ALEXEI B., and M. K. ALI. "LEARNING, EXPLORATION AND CHAOTIC POLICIES." International Journal of Modern Physics C 11, no. 07 (2000): 1455–64. http://dx.doi.org/10.1142/s0129183100001309.
Texto completoResumen
We consider different versions of exploration in reinforcement learning. For the test problem, we use navigation in a shortcut maze. It is shown that chaotic ∊-greedy policy may be as efficient as a random one. The best results were obtained with a model chaotic neuron. Therefore, exploration strategy can be implemented in a deterministic learning system such as a neural network.
6
MORIHIRO, KOICHIRO, NOBUYUKI MATSUI, and HARUHIKO NISHIMURA. "CHAOTIC EXPLORATION EFFECTS ON REINFORCEMENT LEARNING IN SHORTCUT MAZE TASK." International Journal of Bifurcation and Chaos 16, no. 10 (2006): 3015–22. http://dx.doi.org/10.1142/s0218127406016616.
Texto completoResumen
Reinforcement learning is usually required in the process of trial and error called exploration, and the uniform pseudorandom number generator is considered effective in that process. As a generator for the exploration, chaotic sources are also useful in creating a random-like sequence such as in the case of stochastic sources. In this research, we investigate the efficiency of the deterministic chaotic generator for the exploration in learning a nonstationary shortcut maze problem. As a result, it is found that the deterministic chaotic generator based on the logistic map is better in the performance of the exploration than in the stochastic random generator. This has been made clear by analyzing the difference of the performances between the two generators in terms of the patterns of exploration occurrence. We also examine the tent map, which is homeomorphic to the logistic map, compared with other generators.
7
Du, Mengnan, Fengxiang He, Na Zou, Dacheng Tao, and Xia Hu. "Shortcut Learning of Large Language Models in Natural Language Understanding." Communications of the ACM 67, no. 1 (2023): 110–20. http://dx.doi.org/10.1145/3596490.
Texto completo
8
HAN, FANG, MARIAN WIERCIGROCH, JIAN-AN FANG, and ZHIJIE WANG. "EXCITEMENT AND SYNCHRONIZATION OF SMALL-WORLD NEURONAL NETWORKS WITH SHORT-TERM SYNAPTIC PLASTICITY." International Journal of Neural Systems 21, no. 05 (2011): 415–25. http://dx.doi.org/10.1142/s0129065711002924.
Texto completoResumen
Excitement and synchronization of electrically and chemically coupled Newman-Watts (NW) small-world neuronal networks with a short-term synaptic plasticity described by a modified Oja learning rule are investigated. For each type of neuronal network, the variation properties of synaptic weights are examined first. Then the effects of the learning rate, the coupling strength and the shortcut-adding probability on excitement and synchronization of the neuronal network are studied. It is shown that the synaptic learning suppresses the over-excitement, helps synchronization for the electrically coupled network but impairs synchronization for the chemically coupled one. Both the introduction of shortcuts and the increase of the coupling strength improve synchronization and they are helpful in increasing the excitement for the chemically coupled network, but have little effect on the excitement of the electrically coupled one.
9
Hu, Ruilin, Yajun Du, Jingrong Hu, and Hui Li. "Cross-community shortcut detection based on network representation learning and structural features." Intelligent Data Analysis 27, no. 3 (2023): 709–32. http://dx.doi.org/10.3233/ida-216513.
Texto completoResumen
As social networks continue to expand, an increasing number of people prefer to use social networks to post their comments and express their feelings, and as a result, the information contained in social networks has grown explosively. The effective extraction of valuable information from social networks has attracted the attention of many researchers. It can mine hidden information from social networks and promote the development of social network structures. At present, many ranking node approaches, such as structural hole spanners and opinion leaders, are widely adopted to extract valuable information and knowledge. However, approaches for analyzing edge influences are seldom considered. In this study, we proposed an edge PageRank to mine shortcuts (these edges without direct mutual friends) that are located among communities and play an important role in the spread of public opinion. We first used a network-embedding algorithm to order the spanners and determine the direction of every edge. Then, we transferred the graphs of social networks into edge graphs according to the ordering. We considered the nodes and edges of the graphs of the social networks as edges and nodes of the edge graphs, respectively. Finally, we improved the PageRank algorithm on the edge graph to obtained the edge ranking and extracted the shortcuts of social networks. The experimental results for five different sizes of social networks, such as email, YouTube, DBLP-L, DBLP-M, and DBLP-S, verify whether the inferred shortcut is indeed more useful for information dissemination, and the utility of three sets of edges inferred by different methods is compared, namely, the edge inferred by ER, the edge inferred by the Jaccard index. The ER approach improves by approximately 10%, 9.9%, and 8.3% on DBLP, YouTube, and Orkut. Our method is more effective than the edge ranked by the Jaccard index.
10
Zhong, Yujie, Xiao Li, Jiangjian Xie, and Junguo Zhang. "A Lightweight Automatic Wildlife Recognition Model Design Method Mitigating Shortcut Learning." Animals 13, no. 5 (2023): 838. http://dx.doi.org/10.3390/ani13050838.
Texto completoResumen
Recognizing wildlife based on camera trap images is challenging due to the complexity of the wild environment. Deep learning is an optional approach to solve this problem. However, the backgrounds of images captured from the same infrared camera trap are rather similar, and shortcut learning of recognition models occurs, resulting in reduced generality and poor recognition model performance. Therefore, this paper proposes a data augmentation strategy that integrates image synthesis (IS) and regional background suppression (RBS) to enrich the background scene and suppress the existing background information. This strategy alleviates the model’s focus on the background, guiding it to focus on the wildlife in order to improve the model’s generality, resulting in better recognition performance. Furthermore, to offer a lightweight recognition model for deep learning-based real-time wildlife monitoring on edge devices, we develop a model compression strategy that combines adaptive pruning and knowledge distillation. Specifically, a student model is built using a genetic algorithm-based pruning technique and adaptive batch normalization (GA-ABN). A mean square error (MSE) loss-based knowledge distillation method is then used to fine-tune the student model so as to generate a lightweight recognition model. The produced lightweight model can reduce the computational effort of wildlife recognition with only a 4.73% loss in accuracy. Extensive experiments have demonstrated the advantages of our method, which is beneficial for real-time wildlife monitoring with edge intelligence.
11
Trivedi, Anusua, Caleb Robinson, Marian Blazes, et al. "Deep learning models for COVID-19 chest x-ray classification: Preventing shortcut learning using feature disentanglement." PLOS ONE 17, no. 10 (2022): e0274098. http://dx.doi.org/10.1371/journal.pone.0274098.
Texto completoResumen
In response to the COVID-19 global pandemic, recent research has proposed creating deep learning based models that use chest radiographs (CXRs) in a variety of clinical tasks to help manage the crisis. However, the size of existing datasets of CXRs from COVID-19+ patients are relatively small, and researchers often pool CXR data from multiple sources, for example, using different x-ray machines in various patient populations under different clinical scenarios. Deep learning models trained on such datasets have been shown to overfit to erroneous features instead of learning pulmonary characteristics in a phenomenon known as shortcut learning. We propose adding feature disentanglement to the training process. This technique forces the models to identify pulmonary features from the images and penalizes them for learning features that can discriminate between the original datasets that the images come from. We find that models trained in this way indeed have better generalization performance on unseen data; in the best case we found that it improved AUC by 0.13 on held out data. We further find that this outperforms masking out non-lung parts of the CXRs and performing histogram equalization, both of which are recently proposed methods for removing biases in CXR datasets.
12
Lao, Mingrui, Nan Pu, Yu Liu, Kai He, Erwin M. Bakker, and Michael S. Lew. "COCA: COllaborative CAusal Regularization for Audio-Visual Question Answering." Proceedings of the AAAI Conference on Artificial Intelligence 37, no. 11 (2023): 12995–3003. http://dx.doi.org/10.1609/aaai.v37i11.26527.
Texto completoResumen
Audio-Visual Question Answering (AVQA) is a sophisticated QA task, which aims at answering textual questions over given video-audio pairs with comprehensive multimodal reasoning. Through detailed causal-graph analyses and careful inspections of their learning processes, we reveal that AVQA models are not only prone to over-exploit prevalent language bias, but also suffer from additional joint-modal biases caused by the shortcut relations between textual-auditory/visual co-occurrences and dominated answers. In this paper, we propose a COllabrative CAusal (COCA) Regularization to remedy this more challenging issue of data biases. Specifically, a novel Bias-centered Causal Regularization (BCR) is proposed to alleviate specific shortcut biases by intervening bias-irrelevant causal effects, and further introspect the predictions of AVQA models in counterfactual and factual scenarios. Based on the fact that the dominated bias impairing model robustness for different samples tends to be different, we introduce a Multi-shortcut Collaborative Debiasing (MCD) to measure how each sample suffers from different biases, and dynamically adjust their debiasing concentration to different shortcut correlations. Extensive experiments demonstrate the effectiveness as well as backbone-agnostic ability of our COCA strategy, and it achieves state-of-the-art performance on the large-scale MUSIC-AVQA dataset.
13
Natsir, Siti Zahra Mulianti, Bibin Rubini, Didit Ardianto, and Nurhaedah Madjid. "Interactive Learning Multimedia: A Shortcut for Boosting Gen-Z’s Digital literacy in Science Classroom." Jurnal Penelitian Pendidikan IPA 8, no. 5 (2022): 2168–75. http://dx.doi.org/10.29303/jppipa.v8i5.1897.
Texto completoResumen
This study aims to develop interactive multimedia learning to sharpen Gen- Z’s digital literacy on the material of substance pressure. The method used is research and development with research instruments including multimedia validation sheets, digital literacy questionnaires, teachers’ response questionnaires, and Gen- Z response questionnaires. The research began with need analysis, then continued with the multimedia design stage. The multimedia design stage produces the first draft of multimedia and research instruments. The first draft of the multimedia was validated at the development stage. The development stage included the validation of multimedia and research instruments, as well as a limited field test. A valid Multimedia was subsequently implemented into science learning. The evaluation stage was carried out by administering a questionnaire on digital literacy questionnaires and teachers’ and students’ response questionnaires. The results showed that learning multimedia was declared feasible based on a content validity value of 100%, and Gen - Z’s digital literacy index obtained a score of 3.30 or was in the medium category. It can be concluded that interactive learning multimedia can be an alternative way in enhancing Gen-Z's digital on-substance pressure materials. This study shows that science learning can contribute to accelerating digital transformation by enhancing youths’ capacity in digital literacy
14
Fathima, Sheeba. "Music Genre Classification using Deep Learning." International Journal for Research in Applied Science and Engineering Technology 9, no. VII (2021): 66–71. http://dx.doi.org/10.22214/ijraset.2021.36087.
Texto completoResumen
Many subjects are affected by digital music production., including music genre prediction. Machine learning techniques were used to classify music genres in this research. Deep neural networks (DNN) have recently been demonstrated to be effective in a variety of classification tasks. Including music genre classification. In this paper, we propose two methods for boosting music genre classification with convolutional neural networks: 1) using a process inspired by residual learning to combine peak- and average pooling to provide more statistical information to higher level neural networks; and 2) To bypass one or more layers, use shortcut connections. To perform classification, the KNN output is fed into another deep neural network. Our preliminary experimental results on the GTZAN data set show that the above two methods, especially the second one, can effectively improve classification accuracy when compared to two different network topologies.
15
Rees, Simon, Megan Bruce, and Steven Bradley. "Utilising data-driven learning in chemistry teaching: A shortcut to improving chemical language comprehension." New Directions in the Teaching of Physical Sciences, no. 10 (June 1, 2014): 12–19. http://dx.doi.org/10.29311/ndtps.v0i10.511.
Texto completoResumen
This article describes the development of the FOCUS project and specific pedagogical strategies to improve understanding of the language of chemistry. The importance of language comprehension skills for success in learning chemistry has recently been highlighted by Pyburn et al. (2013). The FOCUS project has involved the construction of a database of student writings (from foundation to Ph.D. level) to create a corpus that can then be analysed for the occurrence of key words in context. Using the principles of concordance and data-driven learning (where student becomes language researcher) a number of teaching activities have been developed to enhance the understanding of subject-specific language for both home and international students studying on a foundation level Chemistry course.
16
Rees, Simon, Megan Bruce, and Steven Bradley. "Utilising Data-driven Learning in Chemistry Teaching: a Shortcut to Improving Chemical Language Comprehension." New Directions 10, no. 1 (2014): 12–19. http://dx.doi.org/10.11120/ndir.2014.00028.
Texto completo
17
Wilkinson, Anna, Karin Kuenstner, Julia Mueller, and Ludwig Huber. "Social learning in a non-social reptile ( Geochelone carbonaria )." Biology Letters 6, no. 5 (2010): 614–16. http://dx.doi.org/10.1098/rsbl.2010.0092.
Texto completoResumen
The ability to learn from the actions of another is adaptive, as it is a shortcut for acquiring new information. However, the evolutionary origins of this trait are still unclear. There is evidence that group-living mammals, birds, fishes and insects can learn through observation, but this has never been investigated in reptiles. Here, we show that the non-social red-footed tortoise ( Geochelone carbonaria ) can learn from the actions of a conspecific in a detour task; non-observer animals (without a conspecific demonstrator) failed. This result provides the first evidence that a non-social species can use social cues to solve a task that it cannot solve through individual learning, challenging the idea that social learning is an adaptation for social living.
18
Mengue-Topio, Hursula, Yannick Courbois, Emily K. Farran, and Pascal Sockeel. "Route learning and shortcut performance in adults with intellectual disability: A study with virtual environments." Research in Developmental Disabilities 32, no. 1 (2011): 345–52. http://dx.doi.org/10.1016/j.ridd.2010.10.014.
Texto completo
19
Clegg, Benjamin A. "Stimulus-Specific Sequence Representation in Serial Reaction Time Tasks." Quarterly Journal of Experimental Psychology Section A 58, no. 6 (2005): 1087–101. http://dx.doi.org/10.1080/02724980443000485.
Texto completoResumen
Some recent evidence has favoured purely response-based implicit representation of sequences in serial reaction time tasks. Three experiments were conducted using serial reaction time tasks featuring four spatial stimuli mapped in categories to two responses. Deviant items from the expected sequence that required the expected response resulted in increased response latencies. The findings demonstrated a stimulus-specific form of representation that operates in the serial reaction time task. No evidence was found to suggest that the stimulus-specific learning was contingent on explicit knowledge of the sequence. Such stimulus-based learning would be congruent with a shortcut within an information-processing framework and, combined with other research findings, suggests that there are multiple loci for learning effects.
20
Song, Rui, Fausto Giunchiglia, Yingji Li, Mingjie Tian, and Hao Xu. "TACIT: A Target-Agnostic Feature Disentanglement Framework for Cross-Domain Text Classification." Proceedings of the AAAI Conference on Artificial Intelligence 38, no. 17 (2024): 18999–9007. http://dx.doi.org/10.1609/aaai.v38i17.29866.
Texto completoResumen
Cross-domain text classification aims to transfer models from label-rich source domains to label-poor target domains, giving it a wide range of practical applications. Many approaches promote cross-domain generalization by capturing domaininvariant features. However, these methods rely on unlabeled samples provided by the target domains, which renders the model ineffective when the target domain is agnostic. Furthermore, the models are easily disturbed by shortcut learning in the source domain, which also hinders the improvement of domain generalization ability. To solve the aforementioned issues, this paper proposes TACIT, a target domain agnostic feature disentanglement framework which adaptively decouples robust and unrobust features by Variational Auto-Encoders. Additionally, to encourage the separation of unrobust features from robust features, we design a feature distillation task that compels unrobust features to approximate the output of the teacher. The teacher model is trained with a few easy samples that are easy to carry potential unknown shortcuts. Experimental results verify that our framework achieves comparable results to state-of-the-art baselines while utilizing only source domain data.
21
Suparjan, Suparjan, and Nining Ismiyani. "The Use of Tanjungpura University’s e-Learning-Moodle LMS during Online Learning: Problems, Solutions and Continuation." Ta'dib 26, no. 1 (2023): 71. http://dx.doi.org/10.31958/jt.v26i1.7902.
Texto completoResumen
This study aims to investigate the problems, solutions, and opinions about the use of the Tanjungpura University’s e-learning-Moodle LMS as learning media during the covid-19 pandemic by students of the Elementary School Teacher Education Study Program at the Faculty of Teacher Training and Education of Tanjungpura University in Pontianak, West Kalimantan. Respondents in this study were selected based on the characteristics of gender, and semester classifications. The qualitative research approach was employed in this study, utilizing semi-structured interviews where the interview questions underwent a piloting phase. The results of the study showed that the challenges faced by the students in using Moodle LMS included network instability, application errors, application disadvantages and the level of technology literacy which were followed up by shortcut tricks and strategies the respondents came up with to overcome them. In general, respondents gave a positive response to the use of Moodle LMS which can be seen from their preference for Moodle LMS over other LMS’s, claim of its suitability for use in learning, and willingness to recommend it to other potential users. However, all respondents revealed that they prefer face-to-face learning to online learning.
22
Husain, Arshi, and Virendra P. Vishvakarma. "Optimized deterministic multikernel extreme learning machine for classification of COVID-19 chest Xray images." Journal of Information and Optimization Sciences 44, no. 4 (2023): 771–93. http://dx.doi.org/10.47974/jios-1319.
Texto completoResumen
In this paper, a novel technique has been proposed to exploit the capability of residual network (ResNet) deep learning model to extract the features. It is utilized neither in pretrained form nor as a transfer learning model. ResNet uses shortcut connections to create shortcut blocks in order to skip blocks of convolutional layers (residual blocks). These stacked residual blocks significantly increase training effectiveness and address the degradation issue. For the purpose of classification, a multiple kernel learning based deterministic extreme learning machine (MKD-ELM) which uses a linear combination of different base kernels as target kernel function is designed to classify chest Xray images. Multiple kernels are used here to exploit their non-linear mapping capability on heterogeneous data. MKD-ELM is an enhanced classifier, which does not require iterative training of its parameters. The proposed technique has better feature extraction along with non-iterative training, thus it is having very fast training and very good generalization performance. The kernel and regularization parameters that influence how accurate MKD-ELM is at classifying data, are tuned through experimentation. So, an optimization technique called the genetic algorithm (GA) has been utilized to determine the ideal combination of these parameters for improved performance. The performance of the proposed technique is analysed for COVID-19 detection problem using chest Xray (ChXR) images by changing the training set, types of kernels and coefficients used for combining base kernels. The proposed algorithm achieves a 97.27% recognition rate on first dataset which comprises 5,856 images and 99.06% on the second dataset which consists of 13,808 images. A higher recognition rate is attained for these ChXR image datasets, in respect to modern techniques demonstrating the effectiveness of the proposed algorithm.
23
Xu, Chendong, Weigang Wang, Yunwei Zhang, Jie Qin, Shujuan Yu, and Yun Zhang. "An Indoor Localization System Using Residual Learning with Channel State Information." Entropy 23, no. 5 (2021): 574. http://dx.doi.org/10.3390/e23050574.
Texto completoResumen
With the increasing demand of location-based services, neural network (NN)-based intelligent indoor localization has attracted great interest due to its high localization accuracy. However, deep NNs are usually affected by degradation and gradient vanishing. To fill this gap, we propose a novel indoor localization system, including denoising NN and residual network (ResNet), to predict the location of moving object by the channel state information (CSI). In the ResNet, to prevent overfitting, we replace all the residual blocks by the stochastic residual blocks. Specially, we explore the long-range stochastic shortcut connection (LRSSC) to solve the degradation problem and gradient vanishing. To obtain a large receptive field without losing information, we leverage the dilated convolution at the rear of the ResNet. Experimental results are presented to confirm that our system outperforms state-of-the-art methods in a representative indoor environment.
24
Arun, K., and A. Srinagesh. "Multilingual twitter sentiment analysis using machine learning." International Journal of Electrical and Computer Engineering (IJECE) 10, no. 6 (2020): 5992. http://dx.doi.org/10.11591/ijece.v10i6.pp5992-6000.
Texto completoResumen
Twitter sentiment analysis is one of the leading research fields. Most of the researchers were contributed to twitter sentiment analysis in English tweets, but few researchers focus on the multilingual twitter sentiment analysis. Some challenges are hoping for the research solutions in multilingual twitter sentiment analysis. This study presents the implementation of sentiment analysis in multilingual twitter data and improves the data classification up to the adequate level of accuracy. Twitter is the sixth leading social networking site in the world. Active users for twitter in a month are 330 million. People can tweet or re-tweet in their languages and allow users to use emoji’s, abbreviations, contraction words, miss spellings, and shortcut words. The best platform for sentiment analysis is twitter. Multilingual tweets and data sparsity are the two main challenges. In this paper, the MLTSA algorithm gives the solution for these two challenges. MLTSA algorithm divides into two parts. One is detecting and translating non-English tweets into English using natural language processing (NLP). And the second one is an appropriate pre-processing method with NLP support can reduce the data sparsity. The result of the MLTSA with SVM achieves good accuracy by up to 95%.
25
Zheng, Hui, Yizhi Cao, Min Sun, et al. "Mixed Structure with 3D Multi-Shortcut-Link Networks for Hyperspectral Image Classification." Remote Sensing 14, no. 5 (2022): 1230. http://dx.doi.org/10.3390/rs14051230.
Texto completoResumen
A hyperspectral image classification method based on a mixed structure with a 3D multi-shortcut-link network (MSLN) was proposed for the features of few labeled samples, excess noise, and heterogeneous homogeneity of features in hyperspectral images. First, the spatial–spectral joint features of hyperspectral cube data were extracted through 3D convolution operation; then, the deep network was constructed and the 3D MSLN mixed structure was used to fuse shallow representational features and deep abstract features, while the hybrid activation function was utilized to ensure the integrity of nonlinear data. Finally, the global self-adaptive average pooling and L-softmax classifier were introduced to implement the terrain classification of hyperspectral images. The mixed structure proposed in this study could extract multi-channel features with a vast receptive field and reduce the continuous decay of shallow features while improving the utilization of representational features and enhancing the expressiveness of the deep network. The use of the dropout mechanism and L-softmax classifier endowed the learned features with a better generalization property and intraclass cohesion and interclass separation properties. Through experimental comparative analysis of six groups of datasets, the results showed that this method, compared with the existing deep-learning-based hyperspectral image classification methods, could satisfactorily address the issues of degeneration of the deep network and “the same object with distinct spectra, and distinct objects with the same spectrum.” It could also effectively improve the terrain classification accuracy of hyperspectral images, as evinced by the overall classification accuracies of all classes of terrain objects in the six groups of datasets: 97.698%, 98.851%, 99.54%, 97.961%, 97.698%, and 99.138%.
26
Holmberg, Linn. "Right and Wrong Ways of Knowing." 1700-tal: Nordic Journal for Eighteenth-Century Studies 20 (December 20, 2023): 8–33. http://dx.doi.org/10.7557/4.7203.
Texto completoResumen
This article explores how the eighteenth-century ‘dictionary craze’ – the explosive proliferation of alphabetically organized reference works – can be understood as part of a wider conflict of learning. Drawing on a wide mix of sources, I show that dictionaries, more than any other factual genre of the time, challenged established conventions about what constituted right and wrong ways of reading, learning, and ultimately knowing, and that this was a crucial reason for both the controversy and success of the genre. After an overview of early modern norms of learning, the article examines how eighteenth-century disagreements about factual dictionaries challenged, reproduced, and reconfigured older views. By encouraging readers to follow their own curiosity, read in whatever order they liked, form their own opinions, remember temporarily, forget, and return when needed, dictionaries deviated from established ideals of disciplined study and ‘digestive’ reading, which held that ‘true’ knowledge was deeply incorporated in the individual. The dictionary’s claim to be a ‘shortcut’ to learning also fueled discussions about the very meaning of ‘knowing’, and how much the road to learning could be shortened without missing the goal.
27
Sun, Chaoyue, Ruogu Fang, Marco Salemi, Mattia Prosperi, and Brittany Rife Magalis. "DeepDynaForecast: Phylogenetic-informed graph deep learning for epidemic transmission dynamic prediction." PLOS Computational Biology 20, no. 4 (2024): e1011351. http://dx.doi.org/10.1371/journal.pcbi.1011351.
Texto completoResumen
In the midst of an outbreak or sustained epidemic, reliable prediction of transmission risks and patterns of spread is critical to inform public health programs. Projections of transmission growth or decline among specific risk groups can aid in optimizing interventions, particularly when resources are limited. Phylogenetic trees have been widely used in the detection of transmission chains and high-risk populations. Moreover, tree topology and the incorporation of population parameters (phylodynamics) can be useful in reconstructing the evolutionary dynamics of an epidemic across space and time among individuals. We now demonstrate the utility of phylodynamic trees for transmission modeling and forecasting, developing a phylogeny-based deep learning system, referred to as DeepDynaForecast. Our approach leverages a primal-dual graph learning structure with shortcut multi-layer aggregation, which is suited for the early identification and prediction of transmission dynamics in emerging high-risk groups. We demonstrate the accuracy of DeepDynaForecast using simulated outbreak data and the utility of the learned model using empirical, large-scale data from the human immunodeficiency virus epidemic in Florida between 2012 and 2020. Our framework is available as open-source software (MIT license) at github.com/lab-smile/DeepDynaForcast.
28
Thanuja, B. "Machine Learning Based Crime Rate Analysis Using Python." International Journal for Research in Applied Science and Engineering Technology 10, no. 11 (2022): 1312–16. http://dx.doi.org/10.22214/ijraset.2022.47574.
Texto completoResumen
Abstract: Crime is obvious that the rate of crimes were increasing day by day in all societies in world, but we personally do believe that there are a lot which can be done by both the governments and the individuals to reduce the crimes in communities. Crime analysis is a well-organized way of detecting and examining patterns and trends in crime. We should give utmost importance to study the reasons behind the crimes, so that we can prevent various crimes occurring and we can be able to find suitable solutions to prevent them. When people cannot find work, they have all the free time in the world. They think of crimes as a shortcut to obtaining and processing the riches of life, without any hardwork. To my mind, the overwhelming majority of people tend to participate in activities assisting the government to keep the society a safe place for their own families and the others and for all age groups. Our main aim of this project is to distinguish various crimes using clustering techniques based on the occurrences and regularity. In this project, the crime data is classified using the Support Vector Machine, Decision Tree, Random Forest Algorithm. This proposed system can indicate the areas which has more probability of occurring crimes so that we can easily identify the crimes based on the previous history and we can take measures to prevent the occurrences of crimes.
29
Welte, Peter O. "Caveat Examiner: Beware Clever Students." Perceptual and Motor Skills 77, no. 3_suppl (1993): 1213–14. http://dx.doi.org/10.2466/pms.1993.77.3f.1213.
Texto completoResumen
The administration procedures for the Design Sequences subtest from the Detroit Tests of Learning Aptitude (Third Edition) contain an anomaly which makes it possible for astute examinees to discover a significant shortcut to solving the visual memory task. Discovery of this serendipitous “key” would give some examinees a distinct advantage on the subtest. The time needed to locate target designs is shortened thereby decreasing the time interval during which forgetting occurs. Once the key is uncovered, a significant portion of the memory task evaporates. This could yield artificially inflated memory scores for some students. An alternate administration procedure is proposed keeping in mind the problems attendant upon evaluating data obtained with nonstandardized administrations.
30
Pollock, Mica. "Flipping Our Scripts about Undocumented Immigration." Genealogy 4, no. 1 (2020): 29. http://dx.doi.org/10.3390/genealogy4010029.
Texto completoResumen
This critical family history explores a common script about undocumented immigration: that undocumented immigrants unfairly have refused to “stand in line” for official, sanctioned immigration and instead have broken rules that the rest of “our” families have followed. Noting a hole in her knowledge base, the author put herself on a steep learning curve to “clean her lenses”—to learn more information about opportunities past and present, so she could see and discuss the issue more clearly. The author sought new and forgotten information about immigration history, new information about her own family, and details about actual immigration policy. She wrote this piece to share a few script-flipping realizations, in case they can shortcut this journey for others.
31
Petrov, Sergei, Tapan Mukerji, Xin Zhang, and Xinfei Yan. "Shape Carving Methods of Geologic Body Interpretation from Seismic Data Based on Deep Learning." Energies 15, no. 3 (2022): 1064. http://dx.doi.org/10.3390/en15031064.
Texto completoResumen
The task of seismic data interpretation is a time-consuming and uncertain process. Machine learning tools can help to build a shortcut between raw seismic data and reservoir characteristics of interest. Recently, techniques involving convolutional neural networks have started to gain momentum. Convolutional neural networks are particularly efficient at pattern recognition within images, and this is why they are suitable for seismic facies classification and interpretation tasks. We experimented with three different architectures based on convolutional layers and compared them with different synthetic and field datasets in terms of quality of the seismic interpretation results and computational efficiency. The architectures used in our study were three deep fully convolutional architectures: a 3D convolutional network with a fully connected head; a 2D fully convolutional network, and U-Net. We found the U-Net architecture to be both robust and the fastest when performing classification at the prediction stage. The 3D convolutional model with a fully connected head was the slowest, while a fully convolutional model was unstable in its predictions.
32
Mah, Christopher, Hillary Walker, Lena Phalen, Sarah Levine, Sarah W. Beck, and Jaylen Pittman. "Beyond CheatBots: Examining Tensions in Teachers’ and Students’ Perceptions of Cheating and Learning with ChatGPT." Education Sciences 14, no. 5 (2024): 500. http://dx.doi.org/10.3390/educsci14050500.
Texto completoResumen
As artificial intelligence (AI) is increasingly integrated into educational technologies, teachers and students must acquire new forms of AI literacy, including an understanding of responsible use of AI. In this study, we explored tensions in teachers’ and students’ opinions about what constitutes learning and cheating with AI. Using qualitative methods, we asked Pre-K through postsecondary writing teachers (n = 16) and a linguistically diverse group of students (n = 12) to consider examples of how students might use ChatGPT, rank them in order of how much they thought each student learned and cheated, and explain their rankings. Our study yielded three findings. First, teachers and students used similar criteria to determine their rankings. Second, teachers and students arrived at similar conclusions about learning with ChatGPT but different conclusions about cheating. Finally, disagreements centered on four main tensions between (1) using ChatGPT as a shortcut versus as a scaffold; (2) using ChatGPT to generate ideas versus language; (3) getting support from ChatGPT versus analogous support from other sources; and (4) learning from ChatGPT versus learning without. These findings underscore the importance of student voice in co-constructing norms around responsible AI use.
33
Wang, Tong, Yuan Yao, Feng Xu, Miao Xu, Shengwei An, and Ting Wang. "Inspecting Prediction Confidence for Detecting Black-Box Backdoor Attacks." Proceedings of the AAAI Conference on Artificial Intelligence 38, no. 1 (2024): 274–82. http://dx.doi.org/10.1609/aaai.v38i1.27780.
Texto completoResumen
Backdoor attacks have been shown to be a serious security threat against deep learning models, and various defenses have been proposed to detect whether a model is backdoored or not. However, as indicated by a recent black-box attack, existing defenses can be easily bypassed by implanting the backdoor in the frequency domain. To this end, we propose a new defense DTInspector against black-box backdoor attacks, based on a new observation related to the prediction confidence of learning models. That is, to achieve a high attack success rate with a small amount of poisoned data, backdoor attacks usually render a model exhibiting statistically higher prediction confidences on the poisoned samples. We provide both theoretical and empirical evidence for the generality of this observation. DTInspector then carefully examines the prediction confidences of data samples, and decides the existence of backdoor using the shortcut nature of backdoor triggers. Extensive evaluations on six backdoor attacks, four datasets, and three advanced attacking types demonstrate the effectiveness of the proposed defense.
34
Sulistyo, Totok, and Rohmat Fauzi. "Soil Infiltration Rate Prediction using Machine Learning Regression Model: A Case Study on Sepinggan River Basin, Balikpapan, Indonesia." Indonesian Journal on Geoscience 10, no. 3 (2023): 335–47. http://dx.doi.org/10.17014/ijog.10.3.335-347.
Texto completoResumen
The infiltration rate of soil data is important in a wide range of planning, such as city planning, drainage design, landuse planning, flood prediction, flood disaster mitigation, etc. Collecting data of infiltration through in-site direct measurements is time consuming and costly. Indeed, inferring the infiltration rate using available parameters and the fittest model is needed. The model can shortcut the field measurement to get a predicted accurate infiltration rate that is worthy to support vital planning. This research aims to develop a model of infiltration rate based on initial water contents and grain size of soils. The results are three outstanding models based on the Multiple R Squared, Root Mean Square Error (RMSE), and Mean Average Error (MAE). The implication of the fittest model is reducing the cost and time to get the predicted infiltration rate. The field measurements can be skipped by sampling undisturbed soils and laboratory tests. Keywords: infiltration rate, initial water contents, grain size
35
Ramsgaard Thomsen, Mette, Paul Nicholas, Martin Tamke, Sebastian Gatz, Yuliya Sinke, and Gabriella Rossi. "Towards machine learning for architectural fabrication in the age of industry 4.0." International Journal of Architectural Computing 18, no. 4 (2020): 335–52. http://dx.doi.org/10.1177/1478077120948000.
Texto completoResumen
Machine Learning (ML) is opening new perspectives for architectural fabrication, as it holds the potential for the profession to shortcut the currently tedious and costly setup of digital integrated design to fabrication workflows and make these more adaptable. To establish and alter these workflows rapidly becomes a main concern with the advent of Industry 4.0 in building industry. In this article we present two projects, which presents how ML can lead to radical changes in generation of fabrication data and linking these directly to design intent. We investigate two different moments of implementation: linking performance to the generation of fabrication data (KnitCone) and integrating the ability to adapt fabrication data in realtime as response to fabrication processes (Neural-Network Steered Robotic Fabrication). Together they examine how models can employ design information as training data and be trained to by step processes within the digital chain. We detail the advantages and limitations of each experiment, we reflect on core questions and perspectives of ML for architectural fabrication: the nature of data to be used, the capacity of these algorithms to encode complexity and generalize results, their task-specificness versus their adaptability and the tradeoffs of using them with respect to conventional explicit analytical modelling.
36
KURAKAMI, Takeru, Kazuyoshi SOUMA, Takashi MIYAMOTO, Takahiko FURUYA, Jun MAGOME, and Hiroshi ISHIDAIRA. "APPLICATION OF A DEEP-LEARNING METHOD INCLUDING SHORTCUT PATHS TO CORRECT THE PRECIPITATION OUTPUTS OF A NUMERICAL WEATHER PREDICTION MODEL." Journal of Japan Society of Civil Engineers, Ser. G (Environmental Research) 76, no. 5 (2020): I_471—I_478. http://dx.doi.org/10.2208/jscejer.76.5_i_471.
Texto completo
37
López-Cabrera, José Daniel, Rubén Orozco-Morales, Jorge Armando Portal-Díaz, Orlando Lovelle-Enríquez, and Marlén Pérez-Díaz. "Current limitations to identify covid-19 using artificial intelligence with chest x-ray imaging (part ii). The shortcut learning problem." Health and Technology 11, no. 6 (2021): 1331–45. http://dx.doi.org/10.1007/s12553-021-00609-8.
Texto completo
38
Zhao, Yu, Rennong Yang, Guillaume Chevalier, Ximeng Xu, and Zhenxing Zhang. "Deep Residual Bidir-LSTM for Human Activity Recognition Using Wearable Sensors." Mathematical Problems in Engineering 2018 (December 30, 2018): 1–13. http://dx.doi.org/10.1155/2018/7316954.
Texto completoResumen
Human activity recognition (HAR) has become a popular topic in research because of its wide application. With the development of deep learning, new ideas have appeared to address HAR problems. Here, a deep network architecture using residual bidirectional long short-term memory (LSTM) is proposed. The advantages of the new network include that a bidirectional connection can concatenate the positive time direction (forward state) and the negative time direction (backward state). Second, residual connections between stacked cells act as shortcut for gradients, effectively avoiding the gradient vanishing problem. Generally, the proposed network shows improvements on both the temporal (using bidirectional cells) and the spatial (residual connections stacked) dimensions, aiming to enhance the recognition rate. When testing with the Opportunity dataset and the public domain UCI dataset, the accuracy is significantly improved compared with previous results.
39
J, Kamalakannan, and Chandana Mani R K. "ERNet : Enhanced ResNet for classification of breast histopathological images." ELCVIA Electronic Letters on Computer Vision and Image Analysis 22, no. 2 (2024): 53–68. http://dx.doi.org/10.5565/rev/elcvia.1614.
Texto completoResumen
Inspite of expeditious approaches in field of breast cancer, histopathological analysis is considered as gold standard in diagnosis of cancer. Researchers are working tremendously to automate the detection and analysis of breast histology images, which confess in improving the accuracy and also induce the mimisation of processing time. Deep learning models are providing greater contribution in solving several image classification tasks. In this paper we propose a model to classify breast histological images, which is redesigned from existing ResNet architecture that minimises model parameters and increase computational efficiency. This approach uses enhanced ResNet connection instead of identity shortcut connection used in ResNet architecture. We apply our proposed method on BreakHis dataset and achieve an accuracy around 95.92 %. The numerical results show that our proposed approach outperforms the previous methods with respect to sensitivity and accuracy.
40
He, Z., H. He, J. Li, M. A. Chapman, and H. Ding. "A SHORT-CUT CONNECTIONS-BASED NEURAL NETWORK FOR BUILDING EXTRACTION FROM HIGH RESOLUTION ORTHOIMAGERY." International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XLIII-B1-2022 (May 30, 2022): 39–44. http://dx.doi.org/10.5194/isprs-archives-xliii-b1-2022-39-2022.
Texto completoResumen
Abstract. Extracting building footprints utilizing deep learning-based (DL-based) methods for high-resolution remote sensing images is one of the current research interest areas. However, the extraction results suffer from blurred edges, rounded corners and detail loss in general. Hence, this article presents a detail-oriented deep learning network named eU-Net (enhanced U-Net). The method adopted in this study, imagery send into the pre-module, which consists of the Canny edge detector, Principal Component Analysis (PCA) and the inter-band ratio operations, before feeding them into the network. Then, process skips connections used in the network to reduce the loss of details during edge and corner detection. The encoding and decoding modules, in this network, are redesigned to expand the perceptual field with shortcut connections and stacked layers. Finally, a Dropout module is added in the bottom layer of the network to avoid the over-fitting problem. The experimental results indicate that the methods used in this study outperform other commonly used and state-of-the-art methods of FCN-8s, U-net, DeepLabv3 and Fast SCNN.
41
Liu, Yao, Lianru Gao, Chenchao Xiao, Ying Qu, Ke Zheng, and Andrea Marinoni. "Hyperspectral Image Classification Based on a Shuffled Group Convolutional Neural Network with Transfer Learning." Remote Sensing 12, no. 11 (2020): 1780. http://dx.doi.org/10.3390/rs12111780.
Texto completoResumen
Convolutional neural networks (CNNs) have been widely applied in hyperspectral imagery (HSI) classification. However, their classification performance might be limited by the scarcity of labeled data to be used for training and validation. In this paper, we propose a novel lightweight shuffled group convolutional neural network (abbreviated as SG-CNN) to achieve efficient training with a limited training dataset in HSI classification. SG-CNN consists of SG conv units that employ conventional and atrous convolution in different groups, followed by channel shuffle operation and shortcut connection. In this way, SG-CNNs have less trainable parameters, whilst they can still be accurately and efficiently trained with fewer labeled samples. Transfer learning between different HSI datasets is also applied on the SG-CNN to further improve the classification accuracy. To evaluate the effectiveness of SG-CNNs for HSI classification, experiments have been conducted on three public HSI datasets pretrained on HSIs from different sensors. SG-CNNs with different levels of complexity were tested, and their classification results were compared with fine-tuned ShuffleNet2, ResNeXt, and their original counterparts. The experimental results demonstrate that SG-CNNs can achieve competitive classification performance when the amount of labeled data for training is poor, as well as efficiently providing satisfying classification results.
42
Xu, Pengcheng, Zhongyuan Guo, Lei Liang, and Xiaohang Xu. "MSF-Net: Multi-Scale Feature Learning Network for Classification of Surface Defects of Multifarious Sizes." Sensors 21, no. 15 (2021): 5125. http://dx.doi.org/10.3390/s21155125.
Texto completoResumen
In the field of surface defect detection, the scale difference of product surface defects is often huge. The existing defect detection methods based on Convolutional Neural Networks (CNNs) are more inclined to express macro and abstract features, and the ability to express local and small defects is insufficient, resulting in an imbalance of feature expression capabilities. In this paper, a Multi-Scale Feature Learning Network (MSF-Net) based on Dual Module Feature (DMF) extractor is proposed. DMF extractor is mainly composed of optimized Concatenated Rectified Linear Units (CReLUs) and optimized Inception feature extraction modules, which increases the diversity of feature receptive fields while reducing the amount of calculation; the feature maps of the middle layer with different sizes of receptive fields are merged to increase the richness of the receptive fields of the last layer of feature maps; the residual shortcut connections, batch normalization layer and average pooling layer are used to replace the fully connected layer to improve training efficiency, and make the multi-scale feature learning ability more balanced at the same time. Two representative multi-scale defect data sets are used for experiments, and the experimental results verify the advancement and effectiveness of the proposed MSF-Net in the detection of surface defects with multi-scale features.
43
Malik, Sohail Iqbal, Mohanaad Shakir, Abdalla Eldow, and Mohammed Waseem Ashfaque. "Promoting Algorithmic Thinking in an Introductory Programming Course." International Journal of Emerging Technologies in Learning (iJET) 14, no. 01 (2019): 84. http://dx.doi.org/10.3991/ijet.v14i01.9061.
Texto completoResumen
Computer programming is considered as a difficult area of study for novices. One of the reasons is the main focus of the curriculum presented in an introductory programming (IP) course which emphasizes more on the programming knowledge (syntax and semantic) of the programming language. This study introduced a new teaching curriculum in the IP course which focuses on different skills required by the novices. We compared the IP course materials based on the traditional and new approaches against five categories. The result shows that the new approach encourages both the programming knowledge and problem solving strategies, and promotes deep learning. Furthermore, it discourages programming shortcut (Problem statement → Code), and support three-step approach (Problem statement → Solution Plans → Code) in solving a problem statement. The new approach also promotes algorithmic thinking in the IP course by paying equal attention on the problem solving strategies.
44
Devaney, Kirsty. "‘Waiting for the wow factor’: Perspectives on computer technology in classroom composing." Journal of Music, Technology & Education 12, no. 2 (2019): 121–39. http://dx.doi.org/10.1386/jmte_00002_1.
Texto completoResumen
Due to advancements, affordability and increased accessibility of technology, composing using computer technology has become prevalent in English secondary music classrooms. Despite this, there is still little research investigating the use of technology in music classrooms, resulting in teaching and learning approaches going unchallenged. This article explores how and why computer technology is being used for composing in upper secondary school music classrooms in England. Data were collected through a mixed-methodology approach involving five case-study schools and a survey of 112 classroom music teachers in England. Findings outline both positive and negative aspects of using computer technology to compose, such as how it was often perceived as a shortcut; however it can be argued that the computer software encourages a linear approach to composing, and the unrealistic Musical Instrument Digital Interface (MIDI) sounds can be a demoralising factor for students’ creativity.
45
Xu, Yao, and Qin Yu. "Adaptive Weighted Multi-Level Fusion of Multi-Scale Features: A New Approach to Pedestrian Detection." Future Internet 13, no. 2 (2021): 38. http://dx.doi.org/10.3390/fi13020038.
Texto completoResumen
Great achievements have been made in pedestrian detection through deep learning. For detectors based on deep learning, making better use of features has become the key to their detection effect. While current pedestrian detectors have made efforts in feature utilization to improve their detection performance, the feature utilization is still inadequate. To solve the problem of inadequate feature utilization, we proposed the Multi-Level Feature Fusion Module (MFFM) and its Multi-Scale Feature Fusion Unit (MFFU) sub-module, which connect feature maps of the same scale and different scales by using horizontal and vertical connections and shortcut structures. All of these connections are accompanied by weights that can be learned; thus, they can be used as adaptive multi-level and multi-scale feature fusion modules to fuse the best features. Then, we built a complete pedestrian detector, the Adaptive Feature Fusion Detector (AFFDet), which is an anchor-free one-stage pedestrian detector that can make full use of features for detection. As a result, compared with other methods, our method has better performance on the challenging Caltech Pedestrian Detection Benchmark (Caltech) and has quite competitive speed. It is the current state-of-the-art one-stage pedestrian detection method.
46
Xie, Zhousan, Shikui Tu, and Lei Xu. "Multilevel Attention Network with Semi-supervised Domain Adaptation for Drug-Target Prediction." Proceedings of the AAAI Conference on Artificial Intelligence 38, no. 1 (2024): 329–37. http://dx.doi.org/10.1609/aaai.v38i1.27786.
Texto completoResumen
Prediction of drug-target interactions (DTIs) is a crucial step in drug discovery, and deep learning methods have shown great promise on various DTI datasets. However, existing approaches still face several challenges, including limited labeled data, hidden bias issue, and a lack of generalization ability to out-of-domain data. These challenges hinder the model's capacity to learn truly informative interaction features, leading to shortcut learning and inferior predictive performance on novel drug-target pairs. To address these issues, we propose MlanDTI, a semi-supervised domain adaptive multilevel attention network (Mlan) for DTI prediction. We utilize two pre-trained BERT models to acquire bidirectional representations enriched with information from unlabeled data. Then, we introduce a multilevel attention mechanism, enabling the model to learn domain-invariant DTIs at different hierarchical levels. Moreover, we present a simple yet effective semi-supervised pseudo-labeling method to further enhance our model's predictive ability in cross-domain scenarios. Experiments on four datasets show that MlanDTI achieves state-of-the-art performances over other methods under intra-domain settings and outperforms all other approaches under cross-domain settings. The source code is available at https://github.com/CMACH508/MlanDTI.
47
Mohamed, Islam A., Adel Othman, and Mohamed Fathy. "A new approach to improve reservoir modeling via machine learning." Leading Edge 39, no. 3 (2020): 170–75. http://dx.doi.org/10.1190/tle39030170.1.
Texto completoResumen
In highly heterogeneous basins with complex subsurface geology, such as the Nile Delta Basin, accurate prediction of reservoir modeling has been a challenge. Reservoir modeling is a continuous process that begins with field discovery and ends with the last phases of production and abandonment. Currently, the stochastic reservoir modeling method is widely used instead of the traditional deterministic modeling method to consider spatial statistics and uncertainties. However, the modeling workflow is demanding and slow, typically requiring months from the initial model concept to flow simulation. In addition, errors from early model stages become cumulative and are difficult to change retroactively. To overcome these limitations, a new workflow is proposed that implements probabilistic neural network inversion to predict reservoir properties. First, well-log data were conditioned properly to match the seismic data scale. Then, the networks were trained and validated, using the conditioned well-log data and seismic internal/external attributes, to predict water saturation and effective porosity 3D volumes. The resulting volumes were sampled in simulation 3D grids and tested using a blind well test. Subsequently, the permeability was calculated from a porosity-permeability relationship inside the reservoir. Finally, a dynamic simulation project of the field was performed in which the historical field production and pressures were compared to the predicted values. One of the Pliocene deepwater turbidite reservoirs in the offshore Nile Delta was used to demonstrate the proposed approach. The results proved the accuracy of the model in predicting the reservoir properties and honoring the heterogeneity of the reservoir. The new approach represents a shortcut for the seismic-to-simulation process, providing a reliable and fast way of constructing a reservoir model and making the seismic-to-simulation process easier.
48
Zakareya, Salman, Habib Izadkhah, and Jaber Karimpour. "A New Deep-Learning-Based Model for Breast Cancer Diagnosis from Medical Images." Diagnostics 13, no. 11 (2023): 1944. http://dx.doi.org/10.3390/diagnostics13111944.
Texto completoResumen
Breast cancer is one of the most prevalent cancers among women worldwide, and early detection of the disease can be lifesaving. Detecting breast cancer early allows for treatment to begin faster, increasing the chances of a successful outcome. Machine learning helps in the early detection of breast cancer even in places where there is no access to a specialist doctor. The rapid advancement of machine learning, and particularly deep learning, leads to an increase in the medical imaging community’s interest in applying these techniques to improve the accuracy of cancer screening. Most of the data related to diseases is scarce. On the other hand, deep-learning models need much data to learn well. For this reason, the existing deep-learning models on medical images cannot work as well as other images. To overcome this limitation and improve breast cancer classification detection, inspired by two state-of-the-art deep networks, GoogLeNet and residual block, and developing several new features, this paper proposes a new deep model to classify breast cancer. Utilizing adopted granular computing, shortcut connection, two learnable activation functions instead of traditional activation functions, and an attention mechanism is expected to improve the accuracy of diagnosis and consequently decrease the load on doctors. Granular computing can improve diagnosis accuracy by capturing more detailed and fine-grained information about cancer images. The proposed model’s superiority is demonstrated by comparing it to several state-of-the-art deep models and existing works using two case studies. The proposed model achieved an accuracy of 93% and 95% on ultrasound images and breast histopathology images, respectively.
49
He, Anzheng, Zishuo Dong, Hang Zhang, et al. "Automated Pixel-Level Detection of Expansion Joints on Asphalt Pavement Using a Deep-Learning-Based Approach." Structural Control and Health Monitoring 2023 (May 23, 2023): 1–15. http://dx.doi.org/10.1155/2023/7552337.
Texto completoResumen
Pixel-level detection of expansion joints on complex pavements is significant for traffic safety and the structural integrity of highway bridges. This paper proposed an improved HRNet-OCR, named as expansion joints segmentation network (EJSNet), for automated pixel-level detection of the expansion joints on asphalt pavement. Different from the high-resolution network (HRNet), the proposed EJSNet modifies the residual structure of the first stage by conducting a Conv. + BN + ReLU (convolution + batch normalization + rectified linear unit) operation for each shortcut connection, which can avoid the network degradation. The feature selection module (FSM) and receptive field block (RFB) module are incorporated into the proposed EJSNet model to learn and extract the contexts at different resolution levels for enhanced latent representations. The convolutional block attention module (CBAM) is introduced to enhance the adaptive feature refinement of the network. Moreover, the shared multilayer perceptron (MLP) architecture of the channel attention module (CAM) is also modified in this paper. Experimental results demonstrate that the F-measure and intersection-over-union (IOU) attained by the proposed EJSNet model on 500 testing image sets are 95.14% and 0.9036, respectively. Compared with four state-of-the-art models for semantic segmentation (i.e., SegNet, DeepLabv3+, dual attention network (DANet), and HRNet-OCR), the proposed EJSNet model can yield higher detection accuracy on both private and public datasets.
50
Zhang, Xuetao, Kuangang Fan, Haonan Hou, and Chuankai Liu. "Real-Time Detection of Drones Using Channel and Layer Pruning, Based on the YOLOv3-SPP3 Deep Learning Algorithm." Micromachines 13, no. 12 (2022): 2199. http://dx.doi.org/10.3390/mi13122199.
Texto completoResumen
Achieving a real-time and accurate detection of drones in natural environments is essential for the interception of drones intruding into high-security areas. However, a rapid and accurate detection of drones is difficult because of their small size and fast speed. In this paper a drone detection method as proposed by pruning the convolutional channel and residual structures of YOLOv3-SPP3. First, the k-means algorithm was used to cluster label the boxes. Second, the channel and shortcut layer pruning algorithm was used to prune the model. Third, the model was fine tuned to achieve a real-time detection of drones. The experimental results obtained by using the Ubuntu server under the Python 3.6 environment show that the YOLOv3-SPP3 algorithm is better than YOLOV3, Tiny-YOLOv3, CenterNet, SSD300, and faster R-CNN. There is significant compression in the size, the maximum compression factor is 20.1 times, the maximum detection speed is increased by 10.2 times, the maximum map value is increased by 15.2%, and the maximum precision is increased by 16.54%. The proposed algorithm achieves the mAP score of 95.15% and the detection speed of 112 f/s, which can meet the requirements of the real-time detection of UAVs.