Filters








11 Hits in 1.8 sec

Utilizing Adversarial Targeted Attacks to Boost Adversarial Robustness [article]

Uriya Pesso, Koby Bibas, Meir Feder
2021 arXiv   pre-print
Adversarial attacks have been shown to be highly effective at degrading the performance of deep neural networks (DNNs). The most prominent defense is adversarial training, a method for learning a robust model. Nevertheless, adversarial training does not make DNNs immune to adversarial perturbations. We propose a novel solution by adopting the recently suggested Predictive Normalized Maximum Likelihood. Specifically, our defense performs adversarial targeted attacks according to different
more » ... ses, where each hypothesis assumes a specific label for the test sample. Then, by comparing the hypothesis probabilities, we predict the label. Our refinement process corresponds to recent findings of the adversarial subspace properties. We extensively evaluate our approach on 16 adversarial attack benchmarks using ResNet-50, WideResNet-28, and a2-layer ConvNet trained with ImageNet, CIFAR10, and MNIST, showing a significant improvement of up to 5.7%, 3.7%, and 0.6% respectively.
arXiv:2109.01945v1 fatcat:n4q567hubnginjzczx5z4bjpi4

Single Layer Predictive Normalized Maximum Likelihood for Out-of-Distribution Detection [article]

Koby Bibas, Meir Feder, Tal Hassner
2021 arXiv   pre-print
Bibas et al. (2019b) and Bibas and Feder (2021) showed the pNML solution for linear regression.  ...  The pNML was developed before for linear regression (Bibas et al., 2019b) and was evaluated empirically for DNN (Fu and Levine, 2021; Bibas et al., 2019a) .  ... 
arXiv:2110.09246v1 fatcat:uypwdn7idvftxezpb3zv3dgu3i

Balancing Specialization, Generalization, and Compression for Detection and Tracking [article]

Dotan Kaufman, Koby Bibas, Eran Borenstein, Michael Chertok, Tal Hassner
2019 arXiv   pre-print
We propose a method for specializing deep detectors and trackers to restricted settings. Our approach is designed with the following goals in mind: (a) Improving accuracy in restricted domains; (b) preventing overfitting to new domains and forgetting of generalized capabilities; (c) aggressive model compression and acceleration. To this end, we propose a novel loss that balances compression and acceleration of a deep learning model vs. loss of generalization capabilities. We apply our method to
more » ... the existing tracker and detector models. We report detection results on the VIRAT and CAVIAR data sets. These results show our method to offer unprecedented compression rates along with improved detection. We apply our loss for tracker compression at test time, as it processes each video. Our tests on the OTB2015 benchmark show that applying compression during test time actually improves tracking performance.
arXiv:1909.11348v1 fatcat:wyeiokps2nhwxbzdkdagsnvylm

Learning Rotation Invariant Features for Cryogenic Electron Microscopy Image Reconstruction [article]

Koby Bibas, Gili Weiss-Dicker, Dana Cohen, Noa Cahan, Hayit Greenspan
2021 arXiv   pre-print
Cryo-Electron Microscopy (Cryo-EM) is a Nobel prize-winning technology for determining the 3D structure of particles at near-atomic resolution. A fundamental step in the recovering of the 3D single-particle structure is to align its 2D projections; thus, the construction of a canonical representation with a fixed rotation angle is required. Most approaches use discrete clustering which fails to capture the continuous nature of image rotation, others suffer from low-quality image reconstruction.
more » ... We propose a novel method that leverages the recent development in the generative adversarial networks. We introduce an encoder-decoder with a rotation angle classifier. In addition, we utilize a discriminator on the decoder output to minimize the reconstruction error. We demonstrate our approach with the Cryo-EM 5HDB and the rotated MNIST datasets showing substantial improvement over recent methods.
arXiv:2101.03549v1 fatcat:gqbfz62nczbxldcabsddqqwwye

A New Look at an Old Problem: A Universal Learning Approach to Linear Regression [article]

Koby Bibas, Yaniv Fogel, Meir Feder
2019 arXiv   pre-print
Linear regression is a classical paradigm in statistics. A new look at it is provided via the lens of universal learning. In applying universal learning to linear regression the hypotheses class represents the label y∈ R as a linear combination of the feature vector x^Tθ where x∈ R^M, within a Gaussian error. The Predictive Normalized Maximum Likelihood (pNML) solution for universal learning of individual data can be expressed analytically in this case, as well as its associated learnability
more » ... sure. Interestingly, the situation where the number of parameters M may even be larger than the number of training samples N can be examined. As expected, in this case learnability cannot be attained in every situation; nevertheless, if the test vector resides mostly in a subspace spanned by the eigenvectors associated with the large eigenvalues of the empirical correlation matrix of the training data, linear regression can generalize despite the fact that it uses an "over-parametrized" model. We demonstrate the results with a simulation of fitting a polynomial to data with a possibly large polynomial degree.
arXiv:1905.04708v1 fatcat:li5a5s74lrahdiwupir6tp3jsm

Deep pNML: Predictive Normalized Maximum Likelihood for Deep Neural Networks [article]

Koby Bibas, Yaniv Fogel, Meir Feder
2020 arXiv   pre-print
The pNML has been derived for several model classes in related works (Fogel and Feder, 2018b) such as the barrier (or 1-D perceptron) model, and in Bibas et al. (2019) for the linear regression problem  ... 
arXiv:1904.12286v2 fatcat:jgcm4iruhvdnla3hwj5r2c5omi

Distribution Free Uncertainty for the Minimum Norm Solution of Over-parameterized Linear Regression [article]

Koby Bibas, Meir Feder
2021 arXiv   pre-print
Bibas et al. (2019b) derived the pNML regret for under-parameterized linear regression. Theorem 2 (Bibas et al. (2019b) ).  ...  These work dealt with 1D barrier (Fogel and Feder, 2018) , linear regression (Bibas et al., 2019b) , and the last layer of DNN (Bibas et al., 2019a) .  ... 
arXiv:2102.07181v2 fatcat:2yzkmnl4kbesphmxfghbcrgtgq

Universal Supervised Learning for Individual Data [article]

Yaniv Fogel, Meir Feder
2018 arXiv   pre-print
Acknowledgments Koby Bibas is acknowledged for discussions and for implementing and analyzing the pNML in various problems, from linear regression to deep neural networks.  ...  The joint work with Koby appears in Bibas et al. (2018a,b). We also acknowledge the discussion with Amichai Painsky regarding section 4, and the related work Painsky and Feder (2018) .  ...  These works are reported in Bibas et al. (2018a,b) . In our opinion it will be interesting to find under what "local" conditions on the model class the pNML regret is small.  ... 
arXiv:1812.09520v1 fatcat:4aezx32m25dxnio6wcpfvtwiue

ENDÜSTRİ 4.0 VE VERİMLİLİK: TÜRK BEYAZ EŞYA SEKTÖRÜNDE KEŞFEDİCİ DURUM ÇALIŞMASI

Kübra ŞİMŞEK DEMİRBAĞ, Nihal YILDIRIM
2021 Verimlilik Dergisi  
Nitekim birçok raporda ve akademik araştırmada da Endüstri 4.0 bir marka olarak anılmaktadır (Glas ve Kleeman, 2016; Huchler, 2017; Bíba, 2018; Germany Trade and Invest, 2018; Kheyfets ve Chernova, 2019  ...  Bir KOBİ olan X'in teknoloji yönetimine ayrılmış bir departmanı bulunmamakta, ancak şirket, Endüstri 4.0 çözümleri üreten kardeş şirketinden destek almaktadır.  ... 
doi:10.51551/verimlilik.988466 fatcat:c6iroz2tvnhirlvnl2i3duw6fa

Zespół dewocjonaliów z wykopalisk na cmentarzu przy kościele pw. św. Barbary na Starym Mieście w Częstochowie

Iwona Młodkowska-Przepiórowska
2018 Acta Universitatis Lodziensis Folia Archaeologica  
, czaszka kobie- W tym miejscu pragnę podziękować Panu dr.  ...  R S N S M V [S M Q] L [I] V [B] -IHS VADE RETRO SATANO NON SVADE MIHI VANA SVNT MALA QVE LIBAS IPSE VENENA BIBAS.  ... 
doi:10.18778/0208-6034.33.11 fatcat:34xpxnpsjndirir7vsusnj6pua

The morphology and phonology of metathesis in Amarasi

Owen Edwards
2017 Morphology  
V α V α C# U-form M-form U-form M-form nima → niim 'five' n-biba → n-biib 'massages' Pbeba → Pbeeb 'palm leaves' n-nena → n-neen 'hears' n-sosa → n-soos 'buys' na-tona → na-toon 'tells' n-nuka → n-nuuk  ...  PnenuP → Pnen~nenuP 'turn' kberoP → kber~beroP 'move' msena → msen~sena 'full, satiated' thoe → tho~hoe 'inundate, bless' Proo → Pro~roo 'far, distant' maPfenaP → maPfen~fenaP 'heavy' taikobi → taikob~kobi  ... 
doi:10.1007/s11525-017-9314-y fatcat:atag2emumffzzfyuyyaurcn6fi