Zobrazeno 1 - 10
of 17
pro vyhledávání: '"Kar, Oğuzhan Fatih"'
Autor:
Bachmann, Roman, Kar, Oğuzhan Fatih, Mizrahi, David, Garjani, Ali, Gao, Mingfei, Griffiths, David, Hu, Jiaming, Dehghan, Afshin, Zamir, Amir
Current multimodal and multitask foundation models like 4M or UnifiedIO show promising results, but in practice their out-of-the-box abilities to accept diverse inputs and perform diverse tasks are limited by the (usually rather small) number of moda
Externí odkaz:
http://arxiv.org/abs/2406.09406
Autor:
Kar, Oğuzhan Fatih, Tonioni, Alessio, Poklukar, Petra, Kulshrestha, Achin, Zamir, Amir, Tombari, Federico
Vision-language models (VLMs) are typically composed of a vision encoder, e.g. CLIP, and a language model (LM) that interprets the encoded features to solve downstream tasks. Despite remarkable progress, VLMs are subject to several shortcomings due t
Externí odkaz:
http://arxiv.org/abs/2404.07204
Autor:
Benoit, Harold, Jiang, Liangze, Atanov, Andrei, Kar, Oğuzhan Fatih, Rigotti, Mattia, Zamir, Amir
Supervised learning datasets may contain multiple cues that explain the training set equally well, i.e., learning any of them would lead to the correct predictions on the training data. However, many of them can be spurious, i.e., lose their predicti
Externí odkaz:
http://arxiv.org/abs/2312.16313
Autor:
Mizrahi, David, Bachmann, Roman, Kar, Oğuzhan Fatih, Yeo, Teresa, Gao, Mingfei, Dehghan, Afshin, Zamir, Amir
Current machine learning models for vision are often highly specialized and limited to a single modality and task. In contrast, recent large language models exhibit a wide range of capabilities, hinting at a possibility for similarly versatile models
Externí odkaz:
http://arxiv.org/abs/2312.06647
We propose a method for adapting neural networks to distribution shifts at test-time. In contrast to training-time robustness mechanisms that attempt to anticipate and counter the shift, we create a closed-loop system and make use of a test-time feed
Externí odkaz:
http://arxiv.org/abs/2309.15762
We introduce a set of image transformations that can be used as corruptions to evaluate the robustness of models as well as data augmentation mechanisms for training neural networks. The primary distinction of the proposed transformations is that, un
Externí odkaz:
http://arxiv.org/abs/2203.01441
We present a method for making neural network predictions robust to shifts from the training data distribution. The proposed method is based on making predictions via a diverse set of cues (called 'middle domains') and ensembling them into one strong
Externí odkaz:
http://arxiv.org/abs/2103.10919
Publikováno v:
IEEE Transactions on Computational Imaging (2021)
Spectral imaging is a fundamental diagnostic technique with widespread application. Conventional spectral imaging approaches have intrinsic limitations on spatial and spectral resolutions due to the physical components they rely on. To overcome these
Externí odkaz:
http://arxiv.org/abs/2008.11625
Autor:
Kar, Oğuzhan Fatih, Oktem, Figen S.
Publikováno v:
Opt. Lett. 44, 4582-4585 (2019)
Compressive spectral imaging enables to reconstruct the entire three-dimensional (3D) spectral cube from a few multiplexed images. Here, we develop a novel compressive spectral imaging technique using diffractive lenses. Our technique uses a coded ap
Externí odkaz:
http://arxiv.org/abs/1903.07987
Akademický článek
Tento výsledek nelze pro nepřihlášené uživatele zobrazit.
K zobrazení výsledku je třeba se přihlásit.
K zobrazení výsledku je třeba se přihlásit.