Graph-based Vision-Language Model for Multi-Modal Retrieval of Fashion Products

Artificial Intelligence is reshaping the field of fashion industry in different ways. E-commerce retailers exploit their data through AI to enhance their search engines, make outfit suggestions and forecast the success of a specific fashion product. However, it is a challenging endeavour as the data they possess is huge, complex and multi-modal. The most common way to search for fashion products online is by matching keywords with phrases in the product's description which are often cluttered, inadequate and differ across collections and sellers. A customer may also browse an online store's taxonomy, although this is time-consuming and doesn't guarantee relevant items. With the advent of Deep Learning architectures, particularly Vision-Language models, ad-hoc solutions have been proposed to model both the product image and description to solve this problems. However, the suggested solutions do not exploit effectively the semantic or syntactic information of these modalities, and the unique qualities and relations of clothing items. In this work of thesis, a novel approach is proposed to address this issues, which aims to model and process images and text descriptions as graphs in order to exploit the relations inside and between each modality and employs specific techniques to extract syntactic and semantic information. The results obtained show promising performances on different tasks when compared to the present state-of-the-art deep learning architectures.

Sviluppo di Retrieval-based Chatbot per l'Italiano con Transformer

L’Intelligenza Artificiale negli ultimi anni sta plasmando il futuro dell’umanità in quasi tutti i settori. È già il motore principale di diverse tecnologie emergenti come i big data, la robotica e l’IoT e continuerà ad agire come innovatore tecnologico nel futuro prossimo. Le recenti scoperte e migliorie sia nel campo dell’hardware che in quello matematico hanno migliorato l’efficienza e ridotto i tempi di esecuzione dei software. È in questo contesto che sta evolvendo anche il Natural Language Processing (NLP), un ramo dell’Intelligenza Artificiale che studia il modo in cui fornire ai computer l'abilità di comprendere un testo scritto o parlato allo stesso modo in cui lo farebbe un essere umano. Le ambiguità che distinguono la lingua naturale dalle altre rendono ardui gli studi in questo settore. Molti dei recenti sviluppi algoritmici su NLP si basano su tecnologie inventate decenni fa. La ricerca in questo settore è quindi in continua evoluzione. Questa tesi si pone l'obiettivo di sviluppare la logica di una chatbot help-desk per un'azienda privata. Lo scopo è, sottoposta una domanda da parte di un utente, restituire la risposta associata presente in una collezione domande-risposte. Il problema che questa tesi affronta è sviluppare un modello di NLP in grado di comprendere il significato semantico delle domande in input, poiché esse possono essere formulate in molteplici modi, preservando il contenuto semantico a discapito della sintassi. A causa delle ridotte dimensioni del dataset italiano proprietario su cui testare il modello chatbot, sono state eseguite molteplici sperimentazioni su un ulteriore dataset italiano con task affine. Attraverso diversi approcci di addestramento, tra cui apprendimento metrico, sono state raggiunte alte accuratezze sulle più comuni metriche di valutazione, confermando le capacità del modello proposto e sviluppato.

Effect Of Aloe Vera Application On The Content And Molecular Arrangement Of Glycosaminoglycans During Calcaneal Tendon Healing.

Although several treatments for tendon lesions have been proposed, successful tendon repair remains a great challenge for orthopedics, especially considering the high incidence of re-rupture of injured tendons. Our aim was to evaluate the pharmacological potential of Aloe vera on the content and arrangement of glycosaminoglycans (GAGs) during tendon healing, which was based on the effectiveness of A. vera on collagen organization previously observed by our group. In rats, a partial calcaneal tendon transection was performed with subsequent topical A. vera application at the injury site. The tendons were treated with A. vera ointment for 7 days and excised on the 7(th) , 14(th) , or 21(st) day post-surgery. Control rats received ointment without A. vera. A higher content of GAGs and a lower amount of dermatan sulfate were detected in the A. vera-treated group on the 14(th) day compared with the control. Also at 14 days post-surgery, a lower dichroic ratio in toluidine blue stained sections was observed in A. vera-treated tendons compared with the control. No differences were observed in the chondroitin-6-sulfate and TGF-β1 levels between the groups, and higher amount of non-collagenous proteins was detected in the A. vera-treated group on the 21(st) day, compared with the control group. No differences were observed in the number of fibroblasts, inflammatory cells and blood vessels between the groups. The application of A. vera during tendon healing modified the arrangement of GAGs and increased the content of GAGs and non-collagenous proteins.

Ssr-based Genetic Diversity And Structure Of Garlic Accessions From Brazil.

Garlic is a spice and a medicinal plant; hence, there is an increasing interest in 'developing' new varieties with different culinary properties or with high content of nutraceutical compounds. Phenotypic traits and dominant molecular markers are predominantly used to evaluate the genetic diversity of garlic clones. However, 24 SSR markers (codominant) specific for garlic are available in the literature, fostering germplasm researches. In this study, we genotyped 130 garlic accessions from Brazil and abroad using 17 polymorphic SSR markers to assess the genetic diversity and structure. This is the first attempt to evaluate a large set of accessions maintained by Brazilian institutions. A high level of redundancy was detected in the collection (50 % of the accessions represented eight haplotypes). However, non-redundant accessions presented high genetic diversity. We detected on average five alleles per locus, Shannon index of 1.2, HO of 0.5, and HE of 0.6. A core collection was set with 17 accessions, covering 100 % of the alleles with minimum redundancy. Overall FST and D values indicate a strong genetic structure within accessions. Two major groups identified by both model-based (Bayesian approach) and hierarchical clustering (UPGMA dendrogram) techniques were coherent with the classification of accessions according to maturity time (growth cycle): early-late and midseason accessions. Assessing genetic diversity and structure of garlic collections is the first step towards an efficient management and conservation of accessions in genebanks, as well as to advance future genetic studies and improvement of garlic worldwide.

A novel polar-based human face recognition computational model

Motivated by a recently proposed biologically inspired face recognition approach, we investigated the relation between human behavior and a computational model based on Fourier-Bessel (FB) spatial patterns. We measured human recognition performance of FB filtered face images using an 8-alternative forced-choice method. Test stimuli were generated by converting the images from the spatial to the FB domain, filtering the resulting coefficients with a band-pass filter, and finally taking the inverse FB transformation of the filtered coefficients. The performance of the computational models was tested using a simulation of the psychophysical experiment. In the FB model, face images were first filtered by simulated V1- type neurons and later analyzed globally for their content of FB components. In general, there was a higher human contrast sensitivity to radially than to angularly filtered images, but both functions peaked at the 11.3-16 frequency interval. The FB-based model presented similar behavior with regard to peak position and relative sensitivity, but had a wider frequency band width and a narrower response range. The response pattern of two alternative models, based on local FB analysis and on raw luminance, strongly diverged from the human behavior patterns. These results suggest that human performance can be constrained by the type of information conveyed by polar patterns, and consequently that humans might use FB-like spatial patterns in face processing.

Energy and amino acid content in phase 1 nursery diet: piglet performance and body chemical composition

It was evaluated the effects of metabolizable energy (ME) and digestible lysine (dLYS) densities on performance and body composition of weaned piglets. The study used 114 piglets weaned at 7.4 ± 0.80 kg, out of which 108 were allotted in the nursery and 6 were slaughtered on the weaning day to determine comparative data of body chemical composition. Six nutrients densities were stipulated from a previous study based on the highest nitrogen retention, maintaining the following ME:LYS relationship in the experimental diets: 3,390:1.291; 3,450:1.409; 3,650:1.411; 3,780:1.461; 3,940:1.507; and 4,109 kcal/kg ME:1.564% dLYS. The experimental diets were offered for 13 days when the piglets reached 12.986 ± 1.449 kg of body weight. The probable residual effects of nutritional density on the subsequent performance of the piglets were evaluated. At the end of initial phase 1, six piglets from each density were slaughtered to determine their chemical composition in body fractions and empty body. There was no significant influence of nutritional levels on the performance of the piglets at the end of the evaluation. The results of food conversion and body composition confirm the level indicated in the previous study, 4 g dLYS/Mcal of ME. The increase of energy and lysine densities confirms the need for a correct relationship among both of them to assure better performance of the piglets at the beginning of the growing phase.

Population-based multicentric survey of hepatitis B infection and risk factor differences among three regions in Brazil

This multicentric population-based study in Brazil is the first national effort to estimate the prevalence of hepatitis B (HBV) and risk factors in the capital cities of the Northeast, Central-West, and Federal Districts (2004-2005). Random multistage cluster sampling was used to select persons 13-69 years of age. Markers for HBV were tested by enzyme-linked immunosorbent assay. The HBV genotypes were determined by sequencing hepatitis B surface antigen (HBsAg). Multivariate analyses and simple catalytic model were performed. Overall, 7,881 persons were included; < 70 per cent were not vaccinated. Positivity for HBsAg was less than 1 per cent among non-vaccinated persons and genotypes A, D, and F co-circulated. The incidence of infection increased with age with similar force of infection in all regions. Males and persons having initiated sexual activity were associated with HBV infection in the two settings; healthcare jobs and prior hospitalization were risk factors in the Federal District. Our survey classified these regions as areas with HBV endemicity and highlighted the risk factors differences among the settings

NEAR INFRARED SPECTROSCOPY FOR ESTIMATING SUGARCANE BAGASSE CONTENT IN MEDIUM DENSITY FIBERBOARD

Medium density fiberboard (MDF) is an engineered wood product formed by breaking down selected lignin-cellulosic material residuals into fibers, combining it with wax and a resin binder, and then forming panels by applying high temperature and pressure. Because the raw material in the industrial process is ever-changing, the panel industry requires methods for monitoring the composition of their products. The aim of this study was to estimate the ratio of sugarcane (SC) bagasse to Eucalyptus wood in MDF panels using near infrared (NIR) spectroscopy. Principal component analysis (PCA) and partial least square (PLS) regressions were performed. MDF panels having different bagasse contents were easily distinguished from each other by the PCA of their NIR spectra with clearly different patterns of response. The PLS-R models for SC content of these MDF samples presented a strong coefficient of determination (0.96) between the NIR-predicted and Lab-determined values and a low standard error of prediction (similar to 1.5%) in the cross-validations. A key role of resins (adhesives), cellulose, and lignin for such PLS-R calibrations was shown. PLS-DA model correctly classified ninety-four percent of MDF samples by cross-validations and ninety-eight percent of the panels by independent test set. These NIR-based models can be useful to quickly estimate sugarcane bagasse vs. Eucalyptus wood content ratio in unknown MDF samples and to verify the quality of these engineered wood products in an online process.

INFLUENCE OF ALTITUDE, AGE AND DIAMETER ON YIELD AND ALPHA-BISABOLOL CONTENT OF CANDEIA TREES (Eremanthus erythropappus)

The heartwood of candeia tree is a source of essential oil rich in alpha-bisabolol, a substance widely used in the cosmetic and pharmaceutical industry. Bearing in mind the economic importance of alpha-bisabolol, this work aimed to evaluate the influence of tree age on the yield and content of alpha-bisabolol present in essential oil from candeia, considering two distinct reliefs and three diameter classes, in Aiuruoca region, south Minas Gerais state. The two distinct reliefs correspond respectively to one section of the stand growing at 1,000m of altitude (Area 1) and another section growing at 1,100m of altitude (Area 2). In each section, 15 trees were felled from among 3 different diameter classes. Discs were removed from the base of each tree to estimate their age by doing growth ring count. Soil samples were taken and Subjected to physical and chemical analysis. The logs were reduced into chips and random samples were taken for distillation to extract essential oil. The method used was steam distillation at a pressure of 2 kgf/cm(2)/2.5 h. The chemical analysis was performed in a gas chromatograph (GC) based on the alpha-bisabolol standard reference. The yield of essential oil from trees in Area I was higher than that from trees in Area 2, with the same pattern of influence for older trees. In Area 2, the alpha-bisabolol content was higher in younger trees. No differences were found between the relevant parameters in relation to diameter classes.

Sugarcane genes associated with sucrose content

Background -: Sucrose content is a highly desirable trait in sugarcane as the worldwide demand for cost-effective biofuels surges. Sugarcane cultivars differ in their capacity to accumulate sucrose and breeding programs routinely perform crosses to identify genotypes able to produce more sucrose. Sucrose content in the mature internodes reach around 20% of the culms dry weight. Genotypes in the populations reflect their genetic program and may display contrasting growth, development, and physiology, all of which affect carbohydrate metabolism. Few studies have profiled gene expression related to sugarcane's sugar content. The identification of signal transduction components and transcription factors that might regulate sugar accumulation is highly desirable if we are to improve this characteristic of sugarcane plants. Results -: We have evaluated thirty genotypes that have different Brix (sugar) levels and identified genes differentially expressed in internodes using cDNA microarrays. These genes were compared to existing gene expression data for sugarcane plants subjected to diverse stress and hormone treatments. The comparisons revealed a strong overlap between the drought and sucrose-content datasets and a limited overlap with ABA signaling. Genes associated with sucrose content were extensively validated by qRT-PCR, which highlighted several protein kinases and transcription factors that are likely to be regulators of sucrose accumulation. The data also indicate that aquaporins, as well as lignin biosynthesis and cell wall metabolism genes, are strongly related to sucrose accumulation. Moreover, sucrose-associated genes were shown to be directly responsive to short term sucrose stimuli, confirming their role in sugar-related pathways. Conclusion -: Gene expression analysis of sugarcane populations contrasting for sucrose content indicated a possible overlap with drought and cell wall metabolism processes and suggested signaling and transcriptional regulators to be used as molecular markers in breeding programs. Transgenic research is necessary to further clarify the role of the genes and define targets useful for sugarcane improvement programs based on transgenic plants.

Continental passive microwave-based rainfall estimation algorithm: Application to the Amazon Basin

This paper presents a new statistical algorithm to estimate rainfall over the Amazon Basin region using the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI). The algorithm relies on empirical relationships derived for different raining-type systems between coincident measurements of surface rainfall rate and 85-GHz polarization-corrected brightness temperature as observed by the precipitation radar (PR) and TMI on board the TRMM satellite. The scheme includes rain/no-rain area delineation (screening) and system-type classification routines for rain retrieval. The algorithm is validated against independent measurements of the TRMM-PR and S-band dual-polarization Doppler radar (S-Pol) surface rainfall data for two different periods. Moreover, the performance of this rainfall estimation technique is evaluated against well-known methods, namely, the TRMM-2A12 [ the Goddard profiling algorithm (GPROF)], the Goddard scattering algorithm (GSCAT), and the National Environmental Satellite, Data, and Information Service (NESDIS) algorithms. The proposed algorithm shows a normalized bias of approximately 23% for both PR and S-Pol ground truth datasets and a mean error of 0.244 mm h(-1) ( PR) and -0.157 mm h(-1)(S-Pol). For rain volume estimates using PR as reference, a correlation coefficient of 0.939 and a normalized bias of 0.039 were found. With respect to rainfall distributions and rain area comparisons, the results showed that the formulation proposed is efficient and compatible with the physics and dynamics of the observed systems over the area of interest. The performance of the other algorithms showed that GSCAT presented low normalized bias for rain areas and rain volume [0.346 ( PR) and 0.361 (S-Pol)], and GPROF showed rainfall distribution similar to that of the PR and S-Pol but with a bimodal distribution. Last, the five algorithms were evaluated during the TRMM-Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) 1999 field campaign to verify the precipitation characteristics observed during the easterly and westerly Amazon wind flow regimes. The proposed algorithm presented a cumulative rainfall distribution similar to the observations during the easterly regime, but it underestimated for the westerly period for rainfall rates above 5 mm h(-1). NESDIS(1) overestimated for both wind regimes but presented the best westerly representation. NESDIS(2), GSCAT, and GPROF underestimated in both regimes, but GPROF was closer to the observations during the easterly flow.

The metal content of bulge field stars from FLAMES-GIRAFFE spectra - I. Stellar parameters and iron abundances

Aims. We determine the iron distribution function (IDF) for bulge field stars, in three different fields along the Galactic minor axis and at latitudes b = -4 degrees, b = -6 degrees, and b = -12 degrees. A fourth field including NGC 6553 is also included in the discussion. Methods. About 800 bulge field K giants were observed with the GIRAFFE spectrograph of FLAMES@VLT at spectral resolution R similar to 20 000. Several of them were observed again with UVES at R similar to 45 000 to insure the accuracy of the measurements. The LTE abundance analysis yielded stellar parameters and iron abundances that allowed us to construct an IDF for the bulge that, for the first time, is based on high-resolution spectroscopy for each individual star. Results. The IDF derived here is centered on solar metallicity, and extends from [Fe/H] similar to -1.5 to [Fe/H] similar to + 0.5. The distribution is asymmetric, with a sharper cutoff on the high-metallicity side, and it is narrower than previously measured. A variation in the mean metallicity along the bulge minor axis is clearly between b = -4 degrees and b = -6 degrees ([Fe/H] decreasing similar to by 0.6 dex per kpc). The field at b = -12 degrees. is consistent with the presence of a gradient, but its quantification is complicated by the higher disk/bulge fraction in this field. Conclusions. Our findings support a scenario in which both infall and outflow were important during the bulge formation, and then suggest the presence of a radial gradient, which poses some challenges to the scenario in which the bulge would result solely from the vertical heating of the bar.

Transiting exoplanets from the CoRoT space mission XIII. CoRoT-13b: a dense hot Jupiter in transit around a star with solar metallicity and super-solar lithium content

We announce the discovery of the transiting planet CoRoT-13b. Ground-based follow-up in CFHT and IAC80 confirmed CoRoT's observations. The mass of the planet was measured with the HARPS spectrograph and the properties of the host star were obtained analyzing HIRES spectra from the Keck telescope. It is a hot Jupiter-like planet with an orbital period of 4.04 days, 1.3 Jupiter masses, 0.9 Jupiter radii, and a density of 2.34 g cm(-3). It orbits a G0V star with T(eff) = 5 945 K, M(*) = 1.09 M(circle dot), R(*) = 1.01 R(circle dot), solar metallicity, a lithium content of +1.45 dex, and an estimated age of between 0.12 and 3.15 Gyr. The lithium abundance of the star is consistent with its effective temperature, activity level, and age range derived from the stellar analysis. The density of the planet is extreme for its mass, implies that heavy elements are present with a mass of between about 140 and 300 M(circle plus).

THE STELLAR CONTENT OF OBSCURED GALACTIC GIANT H II REGIONS. VII. W3

Spectrophotometric distances in the K band have been reported by different authors for a number of obscured Galactic H II regions. Almost 50% of them show large discrepancies compared to the classical method using radial velocities measured in the radio spectral region. In order to provide a crucial test of both methods, we selected a target that does not present particular difficulty for any method and which has been measured by as many techniques as possible. The W3 star-forming complex, located in the Perseus arm, offers a splendid opportunity for such a task. We used the Near-Infrared Integral Field Spectrograph on the Frederick C. Gillett Gemini North telescope to classify candidate ""naked photosphere"" OB stars based on Two Micron All Sky Survey photometry. Two of the targets are revealed to be mid-O-type main-sequence stars leading to a distance of d = 2.20 kpc. This is in excellent agreement with the spectrophotometric distance derived in the optical band (d = 2.18 pc) and with a measurement of the W3 trigonometric parallax (d = 1.95 kpc). Such results confirm that the spectrophotometric distances in the K band are reliable. The radio-derived kinematic distance, on the contrary, gives a distance twice as large (d = 4.2 kpc). This indicates that this region of the Perseus arm does not follow the Galactic rotation curve, and this may also be the case for other H II regions for which discrepancies have been found.

Novel insights into the genomic basis of citrus canker based on the genome sequences of two strains of Xanthomonas fuscans subsp aurantifolii

Background: Citrus canker is a disease that has severe economic impact on the citrus industry worldwide. There are three types of canker, called A, B, and C. The three types have different phenotypes and affect different citrus species. The causative agent for type A is Xanthomonas citri subsp. citri, whose genome sequence was made available in 2002. Xanthomonas fuscans subsp. aurantifolii strain B causes canker B and Xanthomonas fuscans subsp. aurantifolii strain C causes canker C. Results: We have sequenced the genomes of strains B and C to draft status. We have compared their genomic content to X. citri subsp. citri and to other Xanthomonas genomes, with special emphasis on type III secreted effector repertoires. In addition to pthA, already known to be present in all three citrus canker strains, two additional effector genes, xopE3 and xopAI, are also present in all three strains and are both located on the same putative genomic island. These two effector genes, along with one other effector-like gene in the same region, are thus good candidates for being pathogenicity factors on citrus. Numerous gene content differences also exist between the three cankers strains, which can be correlated with their different virulence and host range. Particular attention was placed on the analysis of genes involved in biofilm formation and quorum sensing, type IV secretion, flagellum synthesis and motility, lipopolysacharide synthesis, and on the gene xacPNP, which codes for a natriuretic protein. Conclusion: We have uncovered numerous commonalities and differences in gene content between the genomes of the pathogenic agents causing citrus canker A, B, and C and other Xanthomonas genomes. Molecular genetics can now be employed to determine the role of these genes in plant-microbe interactions. The gained knowledge will be instrumental for improving citrus canker control.