1. Perera, Pramuditha and Morariu, Vlad I. and Jain, Rajiv and Manjunatha, Varun and Wigington, Curtis and Ordonez, Vicente and Patel, Vishal M. (2020) {Generative-Discriminative Feature Representations for Open-Set Recognition}. IEEE, https://ieeexplore.ieee.org/document/9157020/, 11811--11820, jun, 10636919, 978-1-7281-7168-5, :C\:/Users/jjy00/Google Drive/ 재 연/ 논 문/new_openset/build_CVPR/new_reference/Perera_Generative-Discriminative_Feature_Representations_for_Open-Set_Recognition_CVPR_2020_paper.pdf:pdf, 10.1109/CVPR42600.2020.01183, 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), We address the problem of open-set recognition, where the goal is to determine if a given sample belongs to one of the classes used for training a model (known classes). The main challenge in open-set recognition is to disentangle open-set samples that produce high class activations from known-set samples. We propose two techniques to force class activations of open-set samples to be low. First, we train a generative model for all known classes and then augment the input with the representation obtained from the generative model to learn a classifier. This network learns to associate high classification probabilities both when the image content is from the correct class as well as when the input and the reconstructed image are consistent with each other. Second, we use self-supervision to force the network to learn more informative features when assigning class scores to improve separation of classes from each other and from open-set samples. We evaluate the performance of the proposed method with recent open-set recognition works across three datasets, where we obtain state-of-the-art results.
2. Neal, Lawrence and Olson, Matthew and Fern, Xiaoli and Wong, Weng-Keen and Li, Fuxin (2018) {Open set learning with counterfactual images}. http://link.springer.com/10.1007/978-3-030-01231-1_38, 613--628, 16113349, 9783030012304, :C\:/Users/jjy00/AppData/Local/Mendeley Ltd./Mendeley Desktop/Downloaded/Neal et al. - 2018 - Open set learning with counterfactual images.pdf:pdf, 10.1007/978-3-030-01231-1_38, Proc. Eur. Conf. Comput. Vis. (ECCV), In open set recognition, a classifier must label instances of known classes while detecting instances of unknown classes not encountered during training. To detect unknown classes while still generalizing to new instances of existing classes, we introduce a dataset augmentation technique that we call counterfactual image generation. Our approach, based on generative adversarial networks, generates examples that are close to training set examples yet do not belong to any training category. By augmenting training with examples generated by this optimization, we can reformulate open set recognition as classification with one additional class, which includes the set of novel and unknown examples. Our approach outperforms existing open set recognition algorithms on a selection of image classification tasks.
3. Wang, Pingfeng and Youn, Byeng D. and Hu, Chao (2012) {A generic probabilistic framework for structural health prognostics and uncertainty management}. Mech. Syst. Signal Process. 28: 622--637 https://doi.org/10.1016/j.ymssp.2011.10.019, http://dx.doi.org/10.1016/j.ymssp.2011.10.019 https://linkinghub.elsevier.com/retrieve/pii/S088832701100450X, Elsevier, 21389249, apr, Health prognostics,Remaining useful life,Similarity,Sparse bayes learning,Synthesized health index,Uncertainty management, 08883270, 0888-3270, :C\:/Users/jjy00/AppData/Local/Mendeley Ltd./Mendeley Desktop/Downloaded/Wang, Youn, Hu - 2012 - A generic probabilistic framework for structural health prognostics and uncertainty management.pdf:pdf, Structural health prognostics can be broadly applied to various engineered artifacts in an engineered system. However, techniques and methodologies for health prognostics become application-specific. This study thus aims at formulating a generic framework of structural health prognostics, which is composed of four core elements: (i) a generic health index system with synthesized health index (SHI), (ii) a generic offline learning scheme using the sparse Bayes learning (SBL) technique, (iii) a generic online prediction scheme using the similarity-based interpolation (SBI), and (iv) an uncertainty propagation map for the prognostic uncertainty management. The SHI enables the use of heterogeneous sensory signals; the sparseness feature employing only a few neighboring kernel functions enables the real-time prediction of remaining useful lives (RULs) regardless of data size; the SBI predicts the RULs with the background health knowledge obtained under uncertain manufacturing and operation conditions; and the uncertainty propagation map enables the predicted RULs to be loaded with their statistical characteristics. The proposed generic framework of structural health prognostics is thus applicable to different engineered systems and its effectiveness is demonstrated with two cases studies. {\textcopyright} 2011 Elsevier Ltd. All rights reserved.
4. Chawla, N. V. and Bowyer, K. W. and Hall, L. O. and Kegelmeyer, W. P. (2002) {SMOTE: Synthetic minority over-sampling technique}. Journal of Artificial Intelligence Research 16
5. (2) : 321--357 https://doi.org/10.1613/jair.953, https://onlinelibrary.wiley.com/doi/abs/10.1002/eap.2043 https://www.jair.org/index.php/jair/article/view/10302, 31758609, jun, air temperature,forest ecological experiment,forest management,photosynthetically active radiation (PAR),relative humidity,soil moisture,soil temperature,temperate deciduous forests,vapor pressure deficit (VPD), 1076-9757, :C\:/Users/jjy00/AppData/Local/Mendeley Ltd./Mendeley Desktop/Downloaded/Chawla et al. - 2002 - SMOTE Synthetic minority over-sampling technique.pdf:pdf, An approach to the construction of classifiers from imbalanced datasets is described. A dataset is imbalanced if the classification categories are not approximately equally represented. Often real-world data sets are predominately composed of ''normal'' examples with only a small percentage of ''abnormal'' or ''interesting'' examples. It is also the case that the cost of misclassifying an abnormal (interesting) example as a normal example is often much higher than the cost of the reverse error. Under-sampling of the majority (normal) class has been proposed as a good means of increasing the sensitivity of a classifier to the minority class. This paper shows that a combination of our method of over-sampling the minority (abnormal) class and under-sampling the majority (normal) class can achieve better classifier performance (in ROC space) than only under-sampling the majority class. This paper also shows that a combination of our method of over-sampling the minority class and under-sampling the majority class can achieve better classifier performance (in ROC space) than varying the loss ratios in Ripper or class priors in Naive Bayes. Our method of over-sampling the minority class involves creating synthetic minority class examples. Experiments are performed using C4.5, Ripper and a Naive Bayes classifier. The method is evaluated using the area under the Receiver Operating Characteristic curve (AUC) and the ROC convex hull strategy.