Filters








118,357 Hits in 4.5 sec

Structured Differential Learning for Automatic Threshold Setting [article]

Jonathan Connell, Benjamin Herta
2018 arXiv   pre-print
We introduce a technique that can automatically tune the parameters of a rule-based computer vision system comprised of thresholds, combinational logic, and time constants.  ...  We describe the components of the system and the associated supervised learning mechanism.  ...  Structured Differential Learning for Automatic Threshold Setting Jonathan H. Connell, Benjamin W. Herta IBM T.J.  ... 
arXiv:1808.00361v1 fatcat:bclkdigtxfh2tbzis3ey5kzcxm

How to do Physics-based Learning [article]

Michael Kellman, Michael Lustig, Laura Waller
2020 arXiv   pre-print
Specifically, we advocate exploiting the auto-differentiation functionality twice, once to build a physics-based network and again to perform physics-based learning.  ...  The goal of this tutorial is to explain step-by-step how to implement physics-based learning for the rapid prototyping of a computational imaging system.  ...  Soft thresholding is used as the proximal operator. the create graph flag of the automatic differentiator to True.  ... 
arXiv:2005.13531v2 fatcat:3xrzhgu42fel3dqdpg2ilnmvmq

Embedding Differentiable Sparsity into Deep Neural Network [article]

Yongjin Lee
2020 arXiv   pre-print
Thus, it can learn the sparsified structure and the weights of networks simultaneously. The proposed approach can learn structured as well as unstructured sparsity.  ...  In this paper, we propose embedding sparsity into the structure of deep neural networks, where model parameters can be exactly zero during training with the stochastic gradient descent.  ...  Coarse Gradient for Threshold Operation If the magnitude of a parameter or a group is less than a threshold, it is zeroed-out by relu and it does not receive a learning signal since the gradient of relu  ... 
arXiv:2006.13716v1 fatcat:xu64fnwz2ndktdhkge2rssquxq

Automatic Clustering Using FSDE-Forced Strategy Differential Evolution

A Yasid
2018 Journal of Physics, Conference Series  
This study intends on acquiring cluster number automatically utilizing forced strategy differential evolution (AC-FSDE).  ...  Clustering analysis is important in datamining for unsupervised data, cause no adequate prior knowledge.  ...  The result shows that automatic clustering using forced strategy of differential evolution (AC-FSDE) more competitive than automatic clustering algorithm based differential evolution (ACDE).  ... 
doi:10.1088/1742-6596/953/1/012127 fatcat:mxcum7io4fevfjjgeu6on6l4cq

Deep Learning Classification for Diabetic Foot Thermograms

Israel Cruz-Vega, Daniel Hernandez-Contreras, Hayde Peregrina-Barreto, Jose de Jesus Rangel-Magdaleno, Juan Manuel Ramirez-Cortes
2020 Sensors  
This paper compares machine learning-based techniques with Deep Learning (DL) structures. We tested common structures in the mode of transfer learning, including AlexNet and GoogleNet.  ...  The main goal of this work is to analyze the use of AI and DL for the classification of diabetic foot thermograms, highlighting their advantages and limitations.  ...  a) RGB image b) Grayscale c) 1 Threshold d) 2 Thresholds e) 3 Thresholds f) 4 Thresholds This automatic segmentation based on DE defines the thresholds of fuzzy sets.  ... 
doi:10.3390/s20061762 pmid:32235780 fatcat:54mue3jv6vdbtar6t5clipym7m

Automatic Interferogram Selection for SBAS-InSAR Based on Deep Convolutional Neural Networks

Yufang He, Guangzong Zhang, Hermann Kaufmann, Guochang Xu
2021 Remote Sensing  
Especially in the era of big data, the demand for an automatic and effective selection method of high-quality interferograms in SBAS-InSAR technology is growing.  ...  It is concluded that DCNN algorithms can automatically select high quality interferogram for the SBAS-InSAR method and thus have a significant impact on the precision of surface deformation monitoring.  ...  Acknowledgments: The authors like to thank the anonymous reviewers for their efforts and constructive comments to improve the quality of this paper.  ... 
doi:10.3390/rs13214468 fatcat:ndto6cjnmjga3lblcqogqkkstu

Automated segmentation technique with self-driven post-processing for histopathological breast cancer images

Chetna Kaushal, Anshu Singla
2020 CAAI Transactions on Intelligence Technology  
area opening, fill holes and selects most appropriate enhanced image required for further analysis.  ...  The post-processing method itself determines the value of different parameters for different operations based on segmented results obtained.  ...  Automated threshold selection reduces the need for initial level set, time step, weighted area coefficient parameters setting.  ... 
doi:10.1049/trit.2019.0077 fatcat:yafdhjlaorgyfccfjspqsjcroi

Neuron segmentation in electron microscopy images using partial differential equations

Cory Jones, Mojtaba Sayedhosseini, Mark Ellisman, Tolga Tasdizen
2013 2013 IEEE 10th International Symposium on Biomedical Imaging  
We also introduce a new method for representing the resulting image that allows for a more dynamic thresholding to further improve the result.  ...  Here we present a partial differential equation with a novel growth term to improve the results of a supervised learning algorithm.  ...  Figure 2 (c) shows that narrow dark structures will have a larger positive value for λ 1 that when subtracted will sharpen membrane structures.  ... 
doi:10.1109/isbi.2013.6556809 pmid:25143802 pmcid:PMC4136503 dblp:conf/isbi/JonesSET13 fatcat:ybvvovdlgzbfrm7aqartkvebwe

Scalable learning of interpretable rules for the dynamic microbiome domain [article]

Venkata Suhas Maringanti, Vanni Bucci, Georg K Gerber
2020 biorxiv/medrxiv   pre-print
We present a new fully-differentiable model that also learns human-interpretable rules for the same classification task, but in an end-to-end gradient-descent based framework.  ...  and moreover learning a larger set of rules, thus providing additional biological insight into the effects of diet and environment on the microbiome.  ...  Differentiable rule learning.  ... 
doi:10.1101/2020.06.25.172270 fatcat:5oxpacwayzhbzlceoc2zwnv3ta

Machine Learning Versus Human-Developed Algorithms in Image Analysis of Microstructures

Adam Piwowarczyk, Leszek Wojnar
2019 Quality Production Improvement - QPI  
Automatically prepared solutions, based on machine learning, constitute an effective and sufficiently precise tool for classification.  ...  Automatic image analysis is nowadays a standard method in quality control of metallic materials, especially in grain size, graphite shape and non-metallic content evaluation.  ...  This is due to the necessity of collecting large data sets for training purposes. The images should be of good quality and represent representative differentiation of microstructures.  ... 
doi:10.2478/cqpi-2019-0056 fatcat:mivgmwsxxrerzowwe2ulxjsbsy

Parameter learning with truncated message-passing

Justin Domke
2011 CVPR 2011  
However, truncated fitting is much faster for learning. Table 1. A summary of the experimental setting.  ...  Namely, one need not store all intermediate values, as in standard automatic differentiation.  ... 
doi:10.1109/cvpr.2011.5995320 dblp:conf/cvpr/Domke11 fatcat:6oky7bc67reijfmqbs4mrtjsza

Latent Patient Network Learning for Automatic Diagnosis [article]

Luca Cosmo, Anees Kazi, Seyed-Ahmad Ahmadi, Nassir Navab, Michael Bronstein
2020 arXiv   pre-print
To this end, we propose a novel, end-to-end trainable graph learning architecture for dynamic and localized graph pruning.  ...  Recently, Graph Convolutional Networks (GCNs) has proven to be a powerful machine learning tool for Computer Aided Diagnosis (CADx) and disease prediction.  ...  Predicting a discrete (e.g. binary) graph structure is a non-differentiable problem.  ... 
arXiv:2003.13620v1 fatcat:cdfxbi5sezaobil4xuc4xrnqum

Differentiable Signal Processing With Black-Box Audio Effects [article]

Marco A. Martínez Ramírez, Oliver Wang, Paris Smaragdis, Nicholas J. Bryan
2021 arXiv   pre-print
commercial solution for music mastering.  ...  To train our network with non-differentiable black-box effects layers, we use a fast, parallel stochastic gradient approximation scheme within a standard auto differentiation graph, yielding efficient  ...  ) [11, 12] , where signal processing structures are implemented within a deep learning auto-differentiation framework and trained via backpropagation.  ... 
arXiv:2105.04752v1 fatcat:j3n5i7q22ndobgpzedaybkt5qa

Focal edge association to glaucoma diagnosis

Jun Cheng, Jiang Liu, Damon Wing Kee Wong, Ngan Meng Tan, Beng Hai Lee, Carol Cheung, M. Baskaran, Tien Yin Wong, Tin Aung
2011 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society  
The association between focal edges and angle grades is built through machine learning. A modified grading system with three grades is adopted.  ...  Moreover, it can correctly classify 75.0% grade 1 and 77.4% grade 0 for angle closure cases.  ...  For the threshold approach, we can introduce another threshold T 2 to differentiate grade 1 from grade 0. We compute the mean width w as in [4] .  ... 
doi:10.1109/iembs.2011.6091111 pmid:22255334 dblp:conf/embc/ChengLWTLCBWA11 fatcat:pzj7wtiylnbercvg73gnj5tu4m

Abstracts of Current Computer Literature

1968 IEEE transactions on computers  
In this manner the automaton is said to possess a variable structure and the ability to learn.  ...  The equivalence is established through a set of defining equations similar to that of a single-threshold threshold element.  ... 
doi:10.1109/tc.1968.227415 fatcat:gww3av6fqncynbg6czmzjuomxi
« Previous Showing results 1 — 15 out of 118,357 results