27 research outputs found
Hyojun dai toa bunzu. 18 , Hawai shoto hen /
No full textMap of Hawaii published in Japan in 1943.; Also available in an electronic version via the internet at: http://nla.gov.au/nla.map-vn6451628. 880-04 Inset: Shinjuwan oyobi Honoruru fukin -- Hawai shoto fukin (Sandoicchi shoto). Scale 1:9,000,000.880-04 Inset: 1-MO!KB'IC!5#i%[i%Ni%ki%k!0o
No full textCreating novel views from a single image has achieved tremendous strides with advanced autoregressive models, as unseen regions have to be inferred from the visible scene contents. Although recent methods generate high-quality novel views, synthesizing with only one explicit or implicit 3D geometry has a trade-off between two objectives that we call the "seesaw" problem: 1) preserving reprojected contents and 2) completing realistic out-of-view regions. Also, autoregressive models require a considerable computational cost. In this paper, we propose a single-image view synthesis framework for mitigating the seesaw problem while utilizing an efficient non-autoregressive model. Motivated by the characteristics that explicit methods well preserve reprojected pixels and implicit methods complete realistic out-of-view regions, we introduce a loss function to complement two renderers. Our loss function promotes that explicit features improve the reprojected area of implicit features and implicit features improve the out-of-view area of explicit features. With the proposed architecture and loss function, we can alleviate the seesaw problem, outperforming autoregressive-based state-of-the-art methods and generating an image approximate to 100 times faster. We validate the efficiency and effectiveness of our method with experiments on RealEstate10 K and ACID datasets.
Rethinking Training Schedules For Verifiably Robust Networks
No full textNew and stronger adversarial attacks can threaten existing defenses. This possibility highlights the importance of certified defense methods that train deep neural networks with verifiably robust guarantees. A range of certified defense methods has been proposed to train neural networks with verifiably robustness guarantees, among which Interval Bound Propagation (IBP) and CROWN-IBP have been demonstrated to be the most effective. However, we observe that CROWN-IBP and IBP are suffering from Low Epsilon Overfitting (LEO), a problem arising from their training schedule that increases the input perturbation bound. We show that LEO can yield poor results even for a simple linear classifier. We also investigate the evidence of LEO from experiments under conditions of worsening LEO. Based on these observations, we propose a new training strategy, BatchMix, which mixes various input perturbation bounds in a mini-batch to alleviate the LEO problem. Experimental results on MNIST and CIFAR10 datasets show that BatchMix can make the performance of IBP and CROWN-IBP better by mitigating LEO
Fine-Grained Multi-Class Object Counting
No full textMany animal species in the wild are at the risk of extinction. To deal with this situation, ecologists have monitored the population changes of endangered species. However, the current wildlife monitoring method is extremely laborious as the animals are counted manually. Automated counting of animals by species can facilitate this work and further renew the ways for ecological studies. However, to the best of our knowledge, few works and publicly available datasets have been proposed on multi-class object counting which is applicable to counting several animal species. In this paper, we propose a fine-grained multi-class object counting dataset, named KRGRUIDAE, which contains endangered red-crowned crane and white-naped crane in the family Gruidae. We also propose a specialized network for multi-class object counting and line segment density maps, and show their effectiveness by comparing results of existing crowd counting methods on the KR-GRUIDAE dataset
검증 가능하게 강건한 뉴럴 네트워크를 위한 훈련 과정 재고
No full text학위논문(석사) - 한국과학기술원 : 전기및전자공학부, 2022.2,[iii, 37 p. :]Adversarial examples which are imperceptibly crafted by adversarial attacks can fool neural networks. Defense methods for it have been proposed, but new and stronger attacks can threaten existing defenses. This possibility highlights the importance of certified defense methods that train deep neural networks with verifiably robust guarantees. Interval bound propagation (IBP)-based methods have been demonstrated to be most effective for certified defense, However, we observe that these methods are suffered from Low Epsilon Overfitting (LEO), a problem arising from their training schedule which increases the input perturbation bound (). In this paper, we show that LEO can disturb the learning of a simple linear classifier in higher epsilon and investigate the evidence of LEO by experiments. Based on these observations, we propose a new training strategy, BatchMix, which mixes various in a mini-batch to alleviate LEO. Experimental results on MNIST and CIFAR-10 datasets show that BatchMix can improve the performance of IBP-based methods.한국과학기술원 :전기및전자공학부
Japanese Dialect Ideology from Meiji to the Present
No full textThe intent of this study is to examine the trajectory of ideology regarding standard Japanese and dialects from the historical perspective, and also to discuss the cause of the post-war shift of the ideology. Before the war, the government attempted to disseminate hyojun-go aiming at creating a unified Japan in the time when many countries were developing to be nation states after industrial revolution. After the Pacific war, the less strict-sounding term kyotsu-go was more often used, conveying an ideology of democratization. Yet despite the difference in the terms, speaking a common language continues to play a role of unifying the country.
Today there is great interest in regional dialects in Japan. Although kyotsu-go is the common language, most people, especially in urban areas, are familiar with (if not fluent in) kyotsu-go. Due to the development of media and mobilization there are few people who cannot understand kyotsu-go. However, until around the 1970s people were more likely to believe in the superiority of standard Japanese (hyojun-go). Standard language was believed to be superior as a result of language policy that had its origins in Meiji and lasted through WWII. This included education policy that required school children to learn hyojun-go. After the war, in a process of democratization there emerged greater acceptance of language variety: dialect. Thus, there has been a shift in language ideology in Japan, and the people\u27s interests in dialects is one indicator of this. This shift is analyzed here from the perspective of Bourdieu\u27s notion of social and linguistic capital, tying it to policy, historical events and societal change
Bridging Implicit and Explicit Geometric Transformation for Single-Image View Synthesis
No full textCreating novel views from a single image has achieved tremendous strides with
advanced autoregressive models, as unseen regions have to be inferred from the
visible scene contents. Although recent methods generate high-quality novel
views, synthesizing with only one explicit or implicit 3D geometry has a
trade-off between two objectives that we call the "seesaw" problem: 1)
preserving reprojected contents and 2) completing realistic out-of-view
regions. Also, autoregressive models require a considerable computational cost.
In this paper, we propose a single-image view synthesis framework for
mitigating the seesaw problem while utilizing an efficient non-autoregressive
model. Motivated by the characteristics that explicit methods well preserve
reprojected pixels and implicit methods complete realistic out-of-view regions,
we introduce a loss function to complement two renderers. Our loss function
promotes that explicit features improve the reprojected area of implicit
features and implicit features improve the out-of-view area of explicit
features. With the proposed architecture and loss function, we can alleviate
the seesaw problem, outperforming autoregressive-based state-of-the-art methods
and generating an image 100 times faster. We validate the efficiency
and effectiveness of our method with experiments on RealEstate10K and ACID
datasets.Comment: TPAMI 202
Hidden Conditional Adversarial Attacks
No full textDeep neural networks are vulnerable to maliciously crafted inputs called adversarial examples. Research on unprecedented adversarial attacks is significant since it can help strengthen the reliability of neural networks by alarming potential threats against them. However, since existing adversarial attacks disturb models unconditionally, the resulting adversarial examples increase their detectability through statistical observations or human inspection. To tackle this limitation, we propose hidden conditional adversarial attacks whose resultant adversarial examples disturb models only if the input images satisfy attackers’ pre-defined conditions. These hidden conditional adversarial examples have better stealthiness and controllability of their attack ability. Our experimental results on the CIFAR-10 and ImageNet datasets show their effectiveness and raise a serious concern about the vulnerability of CNNs against the novel attacks
Exploiting Doubly Adversarial Examples for Improving Adversarial Robustness
No full textDeep neural networks have shown outstanding performance in various areas, but adversarial examples can easily fool them. Although strong adversarial attacks have defeated diverse adversarial defense methods, adversarial training, which augments training data with adversarial examples, remains an effective defense strategy. To further improve adversarial robustness, this paper exploits adversarial examples of adversarial examples. We observe that these doubly adversarial examples tend to return to the original prediction on the clean images but sometimes drift toward other classes. From this finding, we propose a regularization loss that prevents these drifts, which mitigates the vulnerability against multi-targeted attacks. Experimental results on the CIFAR-10 and CIFAR-100 datasets empirically show that the proposed loss improves adversarial robustness
Denoising Task Difficulty-based Curriculum for Training Diffusion Models
No full textDiffusion-based generative models have emerged as powerful tools in the realm
of generative modeling. Despite extensive research on denoising across various
timesteps and noise levels, a conflict persists regarding the relative
difficulties of the denoising tasks. While various studies argue that lower
timesteps present more challenging tasks, others contend that higher timesteps
are more difficult. To address this conflict, our study undertakes a
comprehensive examination of task difficulties, focusing on convergence
behavior and changes in relative entropy between consecutive probability
distributions across timesteps. Our observational study reveals that denoising
at earlier timesteps poses challenges characterized by slower convergence and
higher relative entropy, indicating increased task difficulty at these lower
timesteps. Building on these observations, we introduce an easy-to-hard
learning scheme, drawing from curriculum learning, to enhance the training
process of diffusion models. By organizing timesteps or noise levels into
clusters and training models with ascending orders of difficulty, we facilitate
an order-aware training regime, progressing from easier to harder denoising
tasks, thereby deviating from the conventional approach of training diffusion
models simultaneously across all timesteps. Our approach leads to improved
performance and faster convergence by leveraging benefits of curriculum
learning, while maintaining orthogonality with existing improvements in
diffusion training techniques. We validate these advantages through
comprehensive experiments in image generation tasks, including unconditional,
class-conditional, and text-to-image generation
