No description available.

Content: The research on Collagen that possesses unique fibre structure are reported frequently. In this paper, the cross image of leather fibre of dried wet blue cowhide embedded with and without epoxy resin were investigated with micro computed topography(MCT). The images obtained by MCT of leather fibre are original status without any damage, while the embedded leather can emerge distortion because the fibre was fixed during the solidifying and immersing of the resin. In this research, 2357 images of leather fibre were investigated on wet blue leather(original fibre) and the same piece of leather embedded by epoxy resin(embedded fibre). The area ratio of the sections from the original fibre and the embedded fibre was examined for each image. The statistic results showed that the mode of area ratio of the original fibre section to the entire fibre section is 75%, and the mode of area ratio of the embedded fibre section to the embedded fibre entire section is only 48%. The mode of the area ratio of the original fibre is obviously higher than the mode of the area ratio of the embedded fibre, that is diverse with the anticipation of fibre swelling caused by resin. The reason might be the expansion of interval space among the fibre filled with epoxy resin, otherwise the conglutination of fibre caused by the evaporation of solvent(acetone used in embedding) in the course of the resion solidifying. Likewise, it can be the adhesion of the tiny fibre with the larger fibre that will diminish the area caculated. The factors will be studied further on embedding to achieve a method with minimum deformation on cross image of fibre. Take-Away: The area ratio of cross section on embedded fibre shrunk comparing with the original fibre, that is out of anticipation of the probable swelling created by resin.

No description available.

<p>Embedding information into a printed image is useful in many aspects, in which reliable channel encoding/decoding systems are crucial due to the information loss and error propagation during transmission. So how to improve the transmission accuracy and control the decoding error rate under a predictable level is always crucial to the channel design.</p><p>The current dissertation aims to discuss the design and performance of a two-dimensional coding method for printed materials – Circular Coding. It is a general two-dimensional coding method that allows data recovery with only a cropped portion of the code, and without the knowledge of the carrier image. While some traditional methods add redundancy bits to extend the length of the original massage length, this method embeds the message into image rows in a repeated and shifted manner with redundancy, then uses the majority votes of the redundant bits for recovery.</p><p>We introduce the encoding and decoding system and investigate the performance of the method for noisy and distorted images. For a given required decoding rate, we model the transmission error and compute the minimum requirement for the number of bit repeats.</p><p>Also, we develop a closed form solution to find the the corresponding cropped-window size that will be used for the encoding and decoding system design.</p><p>Finally, we develop a closed-form formula to predict its decoding success rate in a noisy channel under various transmission noise levels, using probabilistic modeling. The theoretical result is validated with simulations. This result enables the optimal parameter selection in the encoder and decoder system design, and decoding rate prediction with different levels of transmission error.</p><p>We also briefly discuss two other projects: development of print quality troubleshooting tools and text line detection in scanned pages.</p>

Sentiment analysis aims at figuring out the opinions of the users towards a certain service or product. In this research, the aim is at classifying the sentiments of users based on the comments they have posed on Douban movie website. In this thesis, I try two different ways to classify the sentiments: with the first one classifying comments into five classes of ratings from 1 to 5, and with the second one classifying comments into three classes of ratings: negative, neutral and positive. For the latter, the ratings of 1 and 2 are grouped as negative, the ratings of 3 neutral and the ratings of 4 and 5 positive. First, Term Frequency Inverse Document Frequency (TF-IDF) is used as the feature extraction technique for machine learning algorithms. Chi Square and Mutual Information are used for feature selection. The selected features are fed into different machine learning methods: Logistic Regression, Linear SVC, SGD classifier and Multinomial Naive Bayes. The performance of models with feature selection will be compared with the performance of models without feature selection for 5-class classification as well as 3-class classification. Also, fastText and Skip-Gram are used as embedding methods for deep learning algorithms LSTM and BILSTM. FastText will also be used for both embedding as well as being a classifier. The aim is to compare different machine learning and deep learning algorithms using different vectorization methods to see which model performs the best regarding both 5-class and 3-class classification. The two classification strategies will be compared with each other in terms of error analysis. The aim is to figure out the similarities and differences of misclassifications made by two different classification strategies.

Les travaux de cette thèse traitent de l'élaboration d'une méthodologie efficace pour la modélisation des effets parasites dans une carte en large bande de fréquence. La réduction du «Time to Market» pour la conception des produits RF et hyperfréquences nécessite le développement d'une méthode efficace de caractérisation et de modélisation pour mieux prendre en compte les erreurs incluses dans la carte de test.Les principaux résultats concernent les contributions suivantes :- Mise au point des standards de calibrage innovateur pour caractériser et modéliser les effets parasites inhérent au modèle.- Élaboration d'une nouvelle approche basée sur une technique de calibrage TRL et une méthode d’élimination efficace de ces effets.- Application aux dispositifs différentiels lors de l'utilisation du calibrage TRL dans le cas de plusieurs ports.La nouvelle approche proposée pour le calibrage et le de-embedding est appliquée à un dispositif actif qui est actuellement utilisé dans l'industrie. Les résultats de mesure d’un dispositif inclus dans une carte de test ont été comparés à des mesures calibrées à l’aide d’une carte d’évaluation comportant des standards TRL.Cette étude a été prolongée avec le calibrage TRL multi-port pour être utilisé pour la bande large des dispositifs comme les dispositifs différentiels. / The work of this thesis deals with the developing of an efficient methodology for modeling parasitic effects within a broadband board. Reducing “Time to Market” for the design of RF and microwave products necessitates the development of an efficient characterization and modeling methodologies for better calibrating the errors embedded within the test board.Main results concern the following contributions:- Development of an innovative calibration standards to characterize and model the parasitic effects embedded within the model.- Elaboration of a new approach based on a TRL calibration technique and de-embedding method effective to de-embed these effects.- Application on differential devices upon using multi-port TRL calibration.The new proposed approach for calibration and de-embedding is applied to an active device which is being in use in industry nowadays. The measurement result of the device within a load board has been compared to a calibrated measurement using an evaluation board that include TRL standards.This study has been extended with multi-port TRL calibration to be used for large variety of devices like the differential ones.

In this thesis we shall provide the reader with results in the field of classical Lorentz spaces. These spaces have been studied since the 50's and have many applications in partial differential equations and interpolation theory. This work includes five papers. First paper studies the properties of Generalized Gamma spaces. Second paper provides an alternative proof of normability characterization of classical Lorentz spaces. The third paper discus conditions of linearity and quasi-norm property of rearrangement-invariant lattices. The following paper gives a characterization of normability of Gamma spaces. And finally the last paper characterizes the embeddings between Generalized Gamma spaces. Powered by TCPDF (www.tcpdf.org)

Unsupervised word embedding methods are frequently used for natural language processing applications. However, the unsupervised methods overlook known lexical relations that can be of value to capture accurate semantic word relations. This thesis aims to explore if Swedish word embeddings can benefit from prior known linguistic information. Four knowledge graphs extracted from Svenska Akademiens ordlista (SAOL) are incorporated during the training process using the Probabilistic Word Embeddings with Laplacian Priors (PELP) model. The four implemented PELP models are compared with baseline results to evaluate the use of side information. The results suggest that various lexical relations in SAOL are of interest to generate more accurate Swedish word embeddings.

No description available.

This work explores a continuum representation for diffuse layer models, thereby endowing continuum embedding models the ability to capture electrostatic phenomena in the environment such as the existence of electrolyte ions, and the nature of ionic liquids. It introduces a new field-aware continuum model that adjusts the size of the quantum regime per atom based on the distribution of charge in a system. The model accounts for the asymmetric nature of solvent distribution when applied to cations versus anions; it also overcomes the need to parameterize continuum interface models for different charged systems. The continuum representation of cavitation in water does not account for the tendency for water to form a hydrogen bonding network that is broken due to the formation of cavities. This effect is a major contributor to hydrophobic solvation and is an important precondition to the investigation of solvated proteins with continuum embedding. A new model inspired by machine learning advances is trained on molecular dynamics simulations due to the difficulty of isolating the cavitation energy term in experiment. Thermodynamic integration is used to calculate the energy from a step-like repulsive potential from cavities in TIP4P water, cavities ranging from small organic molecules, to small proteins. Predictions from this new model show a small improvement for small molecules and scale much better with respect to the size of the system.