A convolutional neural network (CNN) is proposed to learn multiple useful feature representations for a classification from low level (raw pixels) to high level (object). Convolutional kernels are initialized by the learned filter kernels that come from sparse auto-encoders. Unlike some traditional methods, which divide the feature abstracting and classifier training into two separated processes, a discriminative feature vector and a single multi-class classifier of softmax regression are learned simultaneously during the training process. Based on the learned high-quality feature representation, the classification can be efficiently performed. A real-world case of surface defects on steel sheet, which evaluates the classification performance of the proposed method, is depicted in detail. The experimental results indicate that the proposed method is quite simple, effective and robustness for the classification of surface defects on hot-rolled steel sheet. Keywords: convolutional neural networks, classification, surface defects, steel sheet, convolutional kernels, sparse auto-encoder Konvolucijska nevronska mre`a (CNN) je predlagana za u~enje {tevilnih koristnih predstavitev pri klasifikaciji od nizkega nivoja (grobe slikovne pike) do visokega nivoja (predmet). Konvolucijska jedra so inicializirana z nau~enimi filtrirnimi jedri, ki izhajajo iz redkih samoenkoderjev. Razli~no od nekaterih klasi~nih metod, ki delijo funkcijo abstrakcije in trening klasifikacije v dva lo~ena procesa, se vektor nediskriminativne funkcije v enostavnem ve~razrednem klasifikatorju regresije softmax, u~i hkrati med procesom treninga. Na osnovi nau~ene predstavitve z visoko kvalitetno funkcijo, se lahko klasifikacija u~inkovito izvede. Primer iz resni~nega sveta povr{inske napake na jekleni plo~evini, ki ocenjujejo zmogljivost klasifikacije je prikazan v podrobnostih. Rezultati eksperimentov ka`ejo, da je predlagana metoda razmeroma preprosta, u~inkovita in robustna pri klasifikaciji povr{inskih napak na vro~e valjani jekleni plo~evini.

No abstract

Contouring errors calculated from the tracking errors of the individual servo drives is one of the important factors that affect machining accuracy. Through the precompensation of the tracking errors the contouring error can be reduced. This paper builds the prediction model of the tracking error, based on which iterative pre-compensation scheme is proposed, to achieve the optimum precompensation value without running servo drives repetitively. Simulation and experiment results verify that the tracking error can be predicted accurately by the prediction model and the iterative pre-compensation scheme of the tracking error can reduce the contouring error effectively.

A nanofluidic biosensor based on nanoreplica molding photonic crystal (PC) was proposed. UV epoxy PC was fabricated by nanoreplica molding on a master PC wafer. The nanochannels were sealed between the gratings on the PC surface and a taped layer. The resonance wavelength of PC-based nanofluidic biosensor was used for testing the sealing effect. According to the peak wavelength value of the sensor, an initial label-free experiment was realized with R6g as the analyte. When the PC-based biosensor was illuminated by a monochromatic light source with a specific angle, the resonance wavelength of the sensor will match with the light source and amplified the electromagnetic field. The amplified electromagnetic field was used to enhance the fluorescence excitation result. The enhancement effect was used for enhancing fluorescence excitation and emission when matched with the resonance condition. Alexa Fluor 635 was used as the target dye excited by 637-nm laser source on a configured photonic crystal enhanced fluorescence (PCEF) setup, and an initial PCEF enhancement factor was obtained.Electronic supplementary materialThe online version of this article (doi:10.1186/s11671-016-1644-x) contains supplementary material, which is available to authorized users.

No abstract