Filters








71,877 Hits in 5.7 sec

Image Synthesis with a Single (Robust) Classifier [article]

Shibani Santurkar, Dimitris Tsipras, Brandon Tran, Andrew Ilyas, Logan Engstrom, Aleksander Madry
2019 arXiv   pre-print
In contrast to other state-of-the-art approaches, the toolkit we develop is rather minimal: it uses a single, off-the-shelf classifier for all these tasks.  ...  We show that the basic classification framework alone can be used to tackle some of the most challenging tasks in image synthesis.  ...  using a single (robustly trained) classifier.  ... 
arXiv:1906.09453v2 fatcat:5iubo5kgtrfelhjzhordkhksuy

BIGRoC: Boosting Image Generation via a Robust Classifier [article]

Roy Ganz, Michael Elad
2022 arXiv   pre-print
Our method, termed BIGRoC (Boosting Image Generation via a Robust Classifier), is based on a post-processing procedure via the guidance of a given robust classifier and without a need for additional training  ...  The interest of the machine learning community in image synthesis has grown significantly in recent years, with the introduction of a wide range of deep generative models and means for training them.  ...  This experiment is done with a robust and a non-robust classifiers of the same architecture -ResNet50 (He et al., 2016) .  ... 
arXiv:2108.03702v3 fatcat:4fndttdsl5gv5d32ol3fqyxhay

Are Perceptually-Aligned Gradients a General Property of Robust Classifiers? [article]

Simran Kaur, Jeremy Cohen, Zachary C. Lipton
2019 arXiv   pre-print
Our finding supports the hypothesis that perceptually-aligned gradients may be a general property of robust classifiers.  ...  For a standard convolutional neural network, optimizing over the input pixels to maximize the score of some target class will generally produce a grainy-looking version of the original image.  ...  Each row is a (random) starting image, each column is a (random) target class. See Figures 5-6 in Appendix A for more. Figure 3 : 3 Class-conditional image synthesis using a smoothed NN.  ... 
arXiv:1910.08640v2 fatcat:ja2lonfepzaxxlysn5xikcbm5y

Learning Security Classifiers with Verified Global Robustness Properties [article]

Yizheng Chen, Shiqi Wang, Yue Qin, Xiaojing Liao, Suman Jana, David Wagner
2021 arXiv   pre-print
For example, we train a Twitter spam account classifier to satisfy five global robustness properties, with 5.4% decrease in true positive rate, and 0.1% increase in false positive rate, compared to a baseline  ...  In this paper, we present a framework and tools for training classifiers that satisfy global robustness properties.  ...  Global Robustness. Fischer et al. [25] and Melacci et al. [60] proposed global robustness properties for image classifiers using universally quantified statements.  ... 
arXiv:2105.11363v1 fatcat:apeyqnltarbxzhjgb57kbui4pe

Robust Face Alignment Based on Hierarchical Classifier Network [chapter]

Li Zhang, Haizhou Ai, Shihong Lao
2006 Lecture Notes in Computer Science  
As face detection only gives a rough estimation of face region, one important problem is how to align facial shapes starting from this rough estimation, especially on face images with expression and pose  ...  A multi-layer structure is employed to organize the classifiers, which begins with one classifier at the first layer and gradually refines the localization of feature points by more classifiers in the  ...  Acknowledgments This work is supported mainly by a grant from OMRON Corporation. It is also supported in part by National Science Foundation of China under grant No.60332010.  ... 
doi:10.1007/11754336_1 fatcat:zrnbzgabhzcjlomdcj6q66rzxm

Classified

2008 MRS bulletin  
Successful candidates must have a PhD degree in materials science and engineering, chemistry, chemical engineering, or a closely related area.  ...  Rutgers Department of Materials Science and Engineering seeks one Research Associate and two postdoctoral associates for ceramic synthesis and processing research.  ...  All applications should be submitted electronically as a single PDF document to mete@metu.edu.tr.  ... 
doi:10.1557/mrs2008.251 fatcat:4lm2wrfw5jgbhghd72lefcoupy

A ROBUST APPROACH TO CLASSIFY MICROCALCIFICATION IN DIGITAL MAMMOGRAMS USING CONTOURLET TRANSFORM AND SUPPORT VECTOR MACHINE

Jasmine
2013 American Journal of Engineering and Applied Sciences  
The system classifies the mammogram images as normal or abnormal and the abnormal severity as benign or malignant.  ...  Herein we present a novel approach for classifying microcalcification in digital mammograms using Nonsubsampled Contourlet Transform (NSCT) and Support Vector Machine (SVM).  ...  Ordinary histogram equalization simply uses a single histogram for an entire image.  ... 
doi:10.3844/ajeassp.2013.57.68 fatcat:ab7ct7iw6nft5jlhmwjbttukkq

Robustness-via-Synthesis: Robust Training with Generative Adversarial Perturbations [article]

Inci M. Baytas, Debayan Deb
2021 arXiv   pre-print
This study presents a robust training algorithm where the adversarial perturbations are automatically synthesized from a random vector using a generator network.  ...  The classifier is trained with cross-entropy loss regularized with the optimal transport distance between the representations of the natural and synthesized adversarial samples.  ...  Without with a single-step gradient-based attack compared with the utilizing the input image, the generator can output diverse standard PGD adversarial training [9].  ... 
arXiv:2108.09713v1 fatcat:6t5okgu26rg4zlyiktb5am3wrq

Score-Based Generative Classifiers [article]

Roland S. Zimmermann, Lukas Schott, Yang Song, Benjamin A. Dunn, David A. Klindt
2021 arXiv   pre-print
In this work, we investigate score-based generative models as classifiers for natural images.  ...  Additionally, on natural image datasets, previous results have suggested a trade-off between the likelihood of the data and classification accuracy.  ...  E-abs: extending the analysis-by-synthesis robust classification model to more complex image domains.  ... 
arXiv:2110.00473v2 fatcat:6yer6cgkxnbf7kg2m2xmxjtre4

Classified

2006 MRS bulletin  
With its affiliation with the NSF's Fiber and Film Engineering Research Center (CAEFF), Clemson Apparel Research (CAR), the Center for Optical Materials Science and Engineering Technology (COMSET), the  ...  Candidates must hold a PhD degree in materials science and engineering or related discipline, have demonstrated a distinguished record of prior research accomplishments, and be prepared to develop an innovative  ...  All applications should be submitted electronically as a single PDF document to mete@metu.edu.tr.  ... 
doi:10.1557/mrs2006.220 fatcat:kgdjsvoyk5fojfje3ynokd5td4

Classifying facial actions

G. Donato, M.S. Bartlett, J.C. Hager, P. Ekman, T.J. Sejnowski
1999 IEEE Transactions on Pattern Analysis and Machine Intelligence  
This paper explores and compares techniques for automatically recognizing facial actions in sequences of images.  ...  The results provide converging evidence for the importance of using local filters, high spatial frequencies, and statistical independence for classifying facial actions.  ...  FLD projected each class to a single point.Fig. 7. Image synthesis model for the ICA representation. Á Fig. 8 . 8 Sample ICA basis images. Fig. 9 . 9 (a) Shift-invariant local PCA kernels.  ... 
doi:10.1109/34.799905 pmid:21188284 pmcid:PMC3008166 fatcat:wvprfbr25za2bdqrem2x5vvtki

Generative Max-Mahalanobis Classifiers for Image Classification, Generation and More [article]

Xiulong Yang, Hui Ye, Yang Ye, Xiang Li, Shihao Ji
2021 arXiv   pre-print
This paper therefore investigates an LDA classifier for image classification and generation. In particular, the Max-Mahalanobis Classifier (MMC), a special case of LDA, fits our goal very well.  ...  We hypothesize that generative classifiers, such as Linear Discriminant Analysis (LDA), might be more suitable for image generation since generative classifiers model the data generation process explicitly  ...  As an alternative to the softmax classifier utilized in JEM, GMMC has a well-formulated latent feature distribution, which fits well with the generative process of image synthesis. 2.  ... 
arXiv:2101.00122v4 fatcat:kpahcnlfljcercpuf6hkxfcnri

Detection Method for Classifying Malicious Firmware

David Noever, Samantha E. Miller Noever
2021 International journal of network security and its applications  
This work converts the binary headers of 40,000 firmware examples from bytes into 1024-pixel thumbnail images to train a deep neural network.  ...  To explain how the model makes classification decisions, the research applies traditional statistical methods such as both single and ensembles of decision trees with identifiable pixel or byte values  ...  "An analysis of generative adversarial networks and variants for image synthesis on MNIST dataset."  ... 
doi:10.5121/ijnsa.2021.13601 fatcat:m2uopqqovngdzee4zbndpztocq

Classifier-based constraint acquisition

S. D. Prestwich, E. C. Freuder, B. O'Sullivan, D. Browne
2021 Annals of Mathematics and Artificial Intelligence  
We discuss a wide range of possible new acquisition methods with useful properties inherited from classifiers.  ...  sets, and robust under errors.  ...  The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material.  ... 
doi:10.1007/s10472-021-09736-4 fatcat:mgp6qyjy5jdzbeplxzrxewfhry

Learning Adaptive Classifiers Synthesis for Generalized Few-Shot Learning [article]

Han-Jia Ye, Hexiang Hu, De-Chuan Zhan
2021 arXiv   pre-print
We propose the ClAssifier SynThesis LEarning (CASTLE), a learning framework that learns how to synthesize calibrated few-shot classifiers in addition to the multi-class classifiers of head classes with  ...  As a consequence, ACASTLE can handle GFSL with classes from heterogeneous domains effectively.  ...  We use this as a robust evaluation of each system's GFSL capability.  ... 
arXiv:1906.02944v5 fatcat:gwl55e4cpbdr7cy6bopmxyw5ye
« Previous Showing results 1 — 15 out of 71,877 results