@article{NitzeHeidlerBarthetal.2021,
  author    = {Nitze, Ingmar and Heidler, Konrad and Barth, Sophia and Grosse, Guido},
  title     = {Developing and testing a deep learning approach for mapping retrogressive thaw slumps},
  series = {Remote sensing},
  volume    = {13},
  journal   = {Remote sensing},
  number    = {21},
  publisher = {MDPI},
  address   = {Basel},
  issn      = {2072-4292},
  doi       = {10.3390/rs13214294},
  pages     = {23},
  year      = {2021},
  abstract  = {In a warming Arctic, permafrost-related disturbances, such as retrogressive thaw slumps (RTS), are becoming more abundant and dynamic, with serious implications for permafrost stability and bio-geochemical cycles on local to regional scales. Despite recent advances in the field of earth observation, many of these have remained undetected as RTS are highly dynamic, small, and scattered across the remote permafrost region. Here, we assessed the potential strengths and limitations of using deep learning for the automatic segmentation of RTS using PlanetScope satellite imagery, ArcticDEM and auxiliary datasets. We analyzed the transferability and potential for pan-Arctic upscaling and regional cross-validation, with independent training and validation regions, in six different thaw slump-affected regions in Canada and Russia. We further tested state-of-the-art model architectures (UNet, UNet++, DeepLabv3) and encoder networks to find optimal model configurations for potential upscaling to continental scales. The best deep learning models achieved mixed results from good to very good agreement in four of the six regions (maxIoU: 0.39 to 0.58; Lena River, Horton Delta, Herschel Island, Kolguev Island), while they failed in two regions (Banks Island, Tuktoyaktuk). Of the tested architectures, UNet++ performed the best. The large variance in regional performance highlights the requirement for a sufficient quantity, quality and spatial variability in the training data used for segmenting RTS across diverse permafrost landscapes, in varying environmental conditions. With our highly automated and configurable workflow, we see great potential for the transfer to active RTS clusters (e.g., Peel Plateau) and upscaling to much larger regions.},
  language  = {en}
}
@article{RezaeiNaeppiLippertetal.2020,
  author    = {Rezaei, Mina and N{\"a}ppi, Janne J. and Lippert, Christoph and Meinel, Christoph and Yoshida, Hiroyuki},
  title     = {Generative multi-adversarial network for striking the right balance in abdominal image segmentation},
  series = {International journal of computer assisted radiology and surgery},
  volume    = {15},
  journal   = {International journal of computer assisted radiology and surgery},
  number    = {11},
  publisher = {Springer},
  address   = {Berlin},
  issn      = {1861-6410},
  doi       = {10.1007/s11548-020-02254-4},
  pages     = {1847 -- 1858},
  year      = {2020},
  abstract  = {Purpose: The identification of abnormalities that are relatively rare within otherwise normal anatomy is a major challenge for deep learning in the semantic segmentation of medical images. The small number of samples of the minority classes in the training data makes the learning of optimal classification challenging, while the more frequently occurring samples of the majority class hamper the generalization of the classification boundary between infrequently occurring target objects and classes. In this paper, we developed a novel generative multi-adversarial network, called Ensemble-GAN, for mitigating this class imbalance problem in the semantic segmentation of abdominal images. Method: The Ensemble-GAN framework is composed of a single-generator and a multi-discriminator variant for handling the class imbalance problem to provide a better generalization than existing approaches. The ensemble model aggregates the estimates of multiple models by training from different initializations and losses from various subsets of the training data. The single generator network analyzes the input image as a condition to predict a corresponding semantic segmentation image by use of feedback from the ensemble of discriminator networks. To evaluate the framework, we trained our framework on two public datasets, with different imbalance ratios and imaging modalities: the Chaos 2019 and the LiTS 2017. Result: In terms of the F1 score, the accuracies of the semantic segmentation of healthy spleen, liver, and left and right kidneys were 0.93, 0.96, 0.90 and 0.94, respectively. The overall F1 scores for simultaneous segmentation of the lesions and liver were 0.83 and 0.94, respectively. Conclusion: The proposed Ensemble-GAN framework demonstrated outstanding performance in the semantic segmentation of medical images in comparison with other approaches on popular abdominal imaging benchmarks. The Ensemble-GAN has the potential to segment abdominal images more accurately than human experts.},
  language  = {en}
}
@article{YanZhouShuetal.2012,
  author    = {Yan, Ming and Zhou, Wei and Shu, Hua and Kliegl, Reinhold},
  title     = {Lexical and sublexical semantic preview benefits in chinese reading},
  series = {Journal of experimental psychology : Learning, memory, and cognition},
  volume    = {38},
  journal   = {Journal of experimental psychology : Learning, memory, and cognition},
  number    = {4},
  publisher = {American Psychological Association},
  address   = {Washington},
  issn      = {0278-7393},
  doi       = {10.1037/a0026935},
  pages     = {1069 -- 1075},
  year      = {2012},
  abstract  = {Semantic processing from parafoveal words is an elusive phenomenon in alphabetic languages, but it has been demonstrated only for a restricted set of noncompound Chinese characters. Using the gaze-contingent boundary paradigm, this experiment examined whether parafoveal lexical and sublexical semantic information was extracted from compound preview characters. Results generalized parafoveal semantic processing to this representative set of Chinese characters and extended the parafoveal processing to radical (sublexical) level semantic information extraction. Implications for notions of parafoveal information extraction during Chinese reading are discussed.},
  language  = {en}
}
@article{ZhouKlieglYan2013,
  author    = {Zhou, Wei and Kliegl, Reinhold and Yan, Ming},
  title     = {A validation of parafoveal semantic information extraction in reading Chinese},
  series = {Journal of research in reading : a journal of the United Kingdom Reading Association},
  volume    = {36},
  journal   = {Journal of research in reading : a journal of the United Kingdom Reading Association},
  number    = {2},
  publisher = {Wiley-Blackwell},
  address   = {Hoboken},
  issn      = {0141-0423},
  doi       = {10.1111/j.1467-9817.2013.01556.x},
  pages     = {S51 -- S63},
  year      = {2013},
  abstract  = {Parafoveal semantic processing has recently been well documented in reading Chinese sentences, presumably because of language-specific features. However, because of a large variation of fixation landing positions on pretarget words, some preview words actually were located in foveal vision when readers' eyes landed close to the end of the pretarget words. None of the previous studies has completely ruled out a possibility that the semantic preview effects might mainly arise from these foveally processed preview words. This case, whether previously observed positive evidence for parafoveal semantic processing can still hold, has been called into question. Using linear mixed models, we demonstrate in this study that semantic preview benefit from word N+1 decreased if fixation on pretarget word N was close to the preview. We argue that parafoveal semantic processing is not a consequence of foveally processed preview words.},
  language  = {en}
}