In this study, we present a new sparse-representation-based face-classification algorithm that exploits dynamic dictionary optimization on an extended dictionary using synthesized faces. More ...specifically, given a dictionary consisting of face examples, we first augment the dictionary with a set of virtual faces generated by calculating the image difference of a pair of faces. This results in an extended dictionary with hybrid training samples, which enhances the capacity of the dictionary to represent new samples. Second, to reduce the redundancy of the extended dictionary and improve the classification accuracy, we use a dictionary-optimization method. We truncate the extended dictionary with a more compact structure by discarding the original samples with small contributions to represent a test sample. Finally, we perform sparse-representation-based face classification using the optimized dictionary. Experimental results obtained using the AR and FERRET face datasets demonstrate the superiority of the proposed method in terms of accuracy, especially for small-sample-size problems.
User-generated content provides many opportunities for managers and researchers, but insights are hindered by a lack of consensus on how to extract brand-relevant valence and volume. Marketing ...studies use different sentiment extraction tools (SETs) based on social media volume, top-down language dictionaries and bottom-up machine learning approaches. This paper compares the explanatory and forecasting power of these methods over several years for daily customer mindset metrics obtained from survey data. For 48 brands in diverse industries, vector autoregressive models show that volume metrics explain the most for brand awareness and purchase intent, while bottom-up SETs excel at explaining brand impression, satisfaction and recommendation. Systematic differences yield contingent advice: the most nuanced version of bottom-up SETs (SVM with Neutral) performs best for the search goods for all consumer mind-set metrics but Purchase Intent for which Volume metrics work best. For experienced goods, Volume outperforms SVM with neutral. As processing time and costs increase when moving from volume to top-down to bottom-up sentiment extraction tools, these conditional findings can help managers decide when more detailed analytics are worth the investment.
•Sentiment originating from user generated content from social media platforms such as e.g. Facebook can be used to predict traditional mindset metrics such as e.g. Awareness, Consideration, or Satisfaction.•To extract sentiment from text users have choice among top-down (dictionary based) and bottom-up (machine learning based) approaches.•Choice of sentiment extraction tool for each mindset metric depends on industry and brand factors.•Bottom-up SETs excel at explaining and forecasting the mid-funnel metrics from brand impression to satisfaction.•Systematic differences yield contingent advice: bottom-up SETs work best for strong brands, top-down explains more in purchase intent for weaker brands of experience goods.•Volume-based metrics work best for awareness, impression and satisfaction for stronger brands of experience goods.
lLow-dose CT is vital for radiation risk reduction.lThe noise statistic is complex in low-dose CT images and varies in different local image patches.lThe noise property can be iteratively ...characterized in different patches.lThe distributions of sparse representation coefficients can be adaptively determined.lThe proposed weighted adaptive non-local dictionary method is effective for low-dose CT.
Low-dose computed tomography (LDCT) image reconstruction has been attracting much attention in medical applications because it can reduce the radiation risk. Sometimes, traditional methods are difficult to reconstruct satisfying image quality from low-dose projections. It is a challenging task for LDCT reconstruction with image quality improvement. Recently, various patch-based methods including dictionary learning were developed for LDCT and have achieved promising performance. Most of these patch-based methods assume that the noise follows the uniform Gaussian distribution, while noise is much more complex than Gaussian distribution in practice. In this study, considering the varying statistics of noise in different patches, we develop a weighted adaptive non-local dictionary (WAND) method for LDCT. Concretely, instead of establishing a complex model for the noise distribution in the whole image, we iteratively characterize the noise property in each local patch during the iteration processing. Besides, we also adaptively describe the different distributions of sparse coefficients of each patch to better characterize the sparsity priors of the image. The simulated and realistic experiments have shown that WAND can achieve better image quality in terms of small details preservation and noise suppression.
Phase retrieval (PR) aims to recover the image of interest from the recorded phaseless measurement. Traditional PR algorithms that use hand-crafted priors suffer from low-quality reconstructions at ...low signal to noise ratios (SNRs). Recent efforts overcome this limitation by using deep priors, but existing algorithms ignore structural priors. To remedy this issue, we propose a deep unfolded convolutional dictionary learning with the weighted ℓ1-norm, termed DeepCDL, for PR. By doing so, deep priors and structural priors can be utilized. Concretely, we exploit the weighted ℓ1-norm to formulate a convolutional dictionary learning (CDL)-based minimization problem, and then unfold the corresponding iterative algorithm into a deep network architecture. Moreover, we design a data-driven weight generator to generate crucial weights in the weighted ℓ1-norm from representation coefficients. For the PR task, we first utilize structural priors to formulate a PR minimization problem, and then propose an iterative algorithm to deal with the formulated problem. The proposed DeepCDL method is utilized to solve the convolutional dictionary learning subproblem with the weighted ℓ1-norm, and an inertial epigraph method employing the inertial technique is proposed to tackle the image updating subproblem. Furthermore, the proposed PR iterative algorithm is unfolded into a feed-forward network dubbed as DeepCDL-PR, where DeepCDL serves as a prior module and the unfolded inertial epigraph method acts as an image updating module. Experiments demonstrate that DeepCDL-PR can recover higher-quality images at various noise levels, compared with previous PR algorithms.
•We propose a novel deep unfolded convolutional dictionary learning method.•We elaborate a data-driven weight generator to generate optimal weights.•We build a novel deep unfolded PR network.•Our algorithm achieves better reconstructions than existing PR algorithms.
The study of the epidemiology of delirium in hospitalized patients is challenging. We aimed to identify the presence or absence of delirium from clinical text notes using natural language processing ...(NLP) techniques and machine learning (ML) models.
We developed a delirium predictive model using 942 clinical notes from hospitalized patients with an ICD-10 delirium hospital discharge code. Moreover, we implemented ML models using a) delirium-suggestive words from an expert-defined dictionary or b) free text in clinical notes. Both strategies considered positive and negative delirium-associated words.
At the note level, for the dictionary method, the logistic regression model achieved an area under the receiver-operating curve (AUROC) of 0.917 for positive words and 0.914 for combined positive and negative words. The areas under the precision-recall curve (AUPR) were 0.893 and 0.897, respectively. For the free-text method, the model achieved an AUROC of 0.826 and 0.830 and AUPR of 0.852 and 0.856, respectively.
NLP-based ML models accurately identified the presence of delirium in clinical notes. The dictionary-based method was superior to the free-text method. The use of negative features improved performance in both methods.
Our proposed NLP-based ML model identified delirium in clinical notes. This model could automatically screen millions of notes and facilitate the study of the epidemiology of in-hospital delirium.
•We developed a machine learning model and utilized natural language processing to classify clinical notes.•Positive/negative words associated to Delirium are indicated in clinical notes.•Words are sourced from expert-defined medical dictionaries or free text.•Key terms ‘confused’, ‘delirium’, ‘fluctuating’, ‘agitation’ are identified for dictionary-based method.•Key terms ‘confused’, ‘high’, ‘pain’, 'cells', ‘oriented’ are recognized for free-text method.
Creating a Slovene–French LSP Dictionary for translation purposes. In order to compensate for the lack of lexicographical resources for the French–Slovenian language pair in the field of specialized ...translation, a bilateral project has been set up with the aim of developing an online dictionary to be used as a lexicographical tool and as an aid for writing scientific texts. This article presents a method that allows students to actively participate in the process of creating an online database for storing information on specialized terminology and phraseology. We first present the lexicographical situation in Slovenia, especially for the French–Slovenian language pair, and then the objectives of the project as well as the teaching method aimed at creating comparable corpora and lexicographical resources compiled by the students. Finally, we conclude with a synthesis of the results obtained. This method, which has been used since 2018 in a Masters course, provides excellent results from a practical and pedagogical point of view.
O novom kartografskom rječniku Lapaine, Miljenko; Frančula, Nedjeljko; Jazbec, Ivo-Pavao
Studia lexicographica,
11/2021, Volume:
15, Issue:
28
Journal Article, Paper
Peer reviewed
Open access
Godine 2020. objavljen je
Kartografski rječnik
u izdanju
Hrvatskoga kartografskog društva i Naklade Dominović. Od objavljivanja prvoga i
jedinoga
Višejezičnoga kartografskoga rječnika
u Hrvatskoj ...prošlo je
više od 40 godina. U tom razdoblju mnogo toga novoga dogodilo se u svijetu, što
su autori pokušali obuhvatiti novim rječnikom. U ovom radu opisat ćemo probleme
s kojima smo se susreli u radu na pripremi toga rječnika.
The
Cartographic Dictionary
was published by the Croatian
Cartographic Society and Dominović Publishing last year. More than 40 years
have passed since the publication of the first and only
Multilingual
Cartographic Dictionary
in Croatia. In the dictionary, we have attempted to
cover many changes which occurred in the world during this period. The present
paper describes the circumstances and problems we encountered while preparing
the dictionary.
Directional reflectance microscopy (DRM) is a new optical technique that enables grain orientation mapping in crystalline solids by capturing and analyzing light reflectance signals generated by ...chemically etched surfaces. Currently, orientation indexing by DRM relies on fitting the optical signals to identify user-defined features that carry orientation information. This approach is inevitably error-prone and material-dependent. These shortcomings hinder the adoption of DRM as a universal characterization method in materials science. We propose a new indexing method to improve the robustness and versatility of DRM. Our method relies on building a dictionary of all possible reflectance signals generated by a metal, which we simulate using a physics-based forward model that takes crystal orientation as input. We then compare each measured reflectance signal acquired by DRM to all the entries in the dictionary in search of the best match, and thus the correct crystal orientation. We demonstrate our dictionary indexing DRM (DI-DRM) approach on nickel and aluminum polycrystals, which produce markedly different optical reflectance signals. We find that DI-DRM yields measurements with improved accuracy compared to those enabled by fitting the optical signal on both materials and across all crystal orientations considered. We also show that the measurement error (∼3°) is mildly sensitive to experimental variability, including noise, measurement settings, and sample surface preparation. DI-DRM represents a considerable step forward towards the implementation of DRM as a streamline materials characterization technique.
Display omitted