On Randomized Classification Layers and Their Implications in Natural Language Generation

In natural language generation tasks, a neural language model is used for generating a sequence of words forming a sentence. The topmost weight matrix of the language model, known as the classification layer, can be viewed as a set of vectors, each representing a target word from the target dictionary. The target word vectors, along with the rest of the model parameters, are learned and updated during training. In this paper, we analyze the properties encoded in the target vectors and question the necessity of learning these vectors. We suggest to randomly draw the target vectors and set them as fixed so that no weights updates are being made during training. We show that by excluding the vectors from the optimization, the number of parameters drastically decreases with a marginal effect on the performance. We demonstrate the effectiveness of our method in image-captioning and machine-translation.

Multimodal Weighted Fusion of Transformers for Movie Genre Classification

The Multimodal Transformer showed to be a competitive model for multimodal tasks involving textual, visual and audio signals. However, as more modalities are involved, its late fusion by concatenation starts to have a negative impact on the model's performance. Besides, interpreting model's predictions becomes difficult, as one would have to look at the different attention activation matrices. In order to overcome these shortcomings, we propose to perform late fusion by adding a GMU module, which effectively allows the model to weight modalities at instance level, improving its performance while providing a better interpretabilty mechanism. In the experiments, we compare our proposed model (MulT-GMU) against the original implementation (MulT-Concat) and a SOTA model tested in a movie genre classification dataset. Our approach, MulT-GMU, outperforms both, MulT-Concat and previous SOTA model.

FuzzyBIO: A Proposal for Fuzzy Representation of Discontinuous Entities

Discontinuous entities pose a challenge to named entity recognition (NER). These phenomena occur commonly in the biomedical domain. As a solution, expansions of the BIO representation scheme that can handle these entity types are commonly used (i.e. BIOHD). However, the extra tag types make the NER task more difficult to learn. In this paper we propose an alternative; a fuzzy continuous BIO scheme (FuzzyBIO). We focus on the task of Adverse Drug Response extraction and normalization to compare FuzzyBIO to BIOHD. We find that FuzzyBIO improves recall of NER for two of three data sets and results in a higher percentage of correctly identified disjoint and composite entities for all data sets. Using FuzzyBIO also improves end-to-end performance for continuous and composite entities in two of three data sets. Since FuzzyBIO improves performance for some data sets and the conversion from BIOHD to FuzzyBIO is straightforward, we recommend investigating which is more effective for any data set containing discontinuous entities.

Proposed Model for Estimating the Total Social Demand for Health Services in Lattakia Governorate between 2003 and 2018

The demand for health services is of great importance as it is linked to the social aspect, which is the health status of citizens in a country, and it is from this importance that our current study focusing on a very important problem related to the ability to estimate the total social demand for health services was chosen. The study was the reality of social demand On health services in the public sector in Lattakia Governorate is one of the most important objectives of the study, in addition to studying the reality of social demand for health services in the private sector in Lattakia Governorate, and working on proposing a model for estimating social demand for health services in the public and private sectors.An analytical descriptive approach has been used, and at the end of the study the study reached an important set of results related to the port of the model and the estimation of social demand for health services in pharmacies, clinics and laboratories based on a questionnaire distributed to families in Latakia, in addition to the opinion of specialists from doctors, pharmacists, laboratory owners, and specialists At the Doctors Syndicate, the Dental Association, and the Pharmacists Syndicate. The health insurance system applied to workers in the country also contributed to increasing the demand for medical treatment in private sector clinics, as well as increasing social demand for health services provided in public hospitals during the study period, and the current crisis and the war on Syria have increased the demand for health services in general in Lattakia Governorate.

Relationship between Red Blood Cell Distribution Width (RDW), Bilirubin, and Clinical Characteristics of Patients with Gastric Cancer

The purpose of the present study was to investigate whether a relationship exists between bilirubin, RDW, and prognostic factors for gastric cancer (tumor stage - tumor markers). Methods: The study included 112 individuals admitted to Tishreen University Hospital and the Oncology Center in Latakia during the period between July 2018 and December 2019. They were distributed into two groups as follow: the group of gastric cancer patients included 56 patients (33 males and 23 females) between the ages of 37 and 83 years, and the control group included 56 individuals (34 males and 22 females) between the ages of 37 and 80 years. The stage of gastric cancer was evaluated based on the AJCC (American Joint Committee on Cancer) cancer staging system. Patients were interrogated, and social and behavioral information were obtained, then venous blood samples were obtained for biomedical evaluation. Statistical analysis was done using SPSS version 25 software. Results: Bilirubin and RDW were significantly associated with the prognostic factors of gastric cancer. RDW was positively associated with the tumor stage and tumor markers (CEA and CA19-9). Total bilirubin (TBIL) and indirect bilirubin (IBIL) were negatively associated with these factors, while direct bilirubin (DBIL) showed no significant association. Patients with gastric cancer had higher RDW than healthy controls, while total bilirubin (TBIL) and indirect bilirubin (IBIL) were significantly decreased. The results of our study indicate that both RDW and bilirubin could be potential prognostic factor for gastric cancer.

Evaluate the performance of a wide band LPDA antenna and compare it with a half-wavelength antenna at frequency change

Antennas have become an indispensable basic need, especially in the area of wireless networks, and have developed in the past few decades in a very large way in terms of profit and resistance, bandwidth, bandwidth, radiation pattern, etc. in line with new and increasing requirements. In this study, we are interested in studying a broad band antenna where this antenna is very important in systems that operate on a set of frequencies rather than at a fixed frequency. The traditional antennas that are designed at a specific frequency change their response significantly if used on a wide bandwidth around the designed frequency In this study, the LPDA (Log periodic dipole Array) antenna was designed and studied in a broadband field and compared to the antenna reference of wire antennas. It is a halfwavelength antenna λ/2. The study of the effect of frequency change in the antenna radiation pattern and the radiation beam Half-power level and front-to-back ratio (F/B).

Studying And Analyzing the modulators and demodulators of PCM

Modulating signals of transmission stations to reception stations is a key factor to guarantee the best possible transmission and reception of these signals .Digital modulation represents a huge evolution in communication field and modulation, which used to depend on analog signal modulation of one parameter-Amplitude . frequency or phase. Digital modulation depends on transforming the transmitted data signal (Bits) and then sending it as samples, and changed back into an analog signals in reception station . In digital systems, digital data are transformed into analog data in the transmitter and does the reverse in the receiver. In digital transmission, on the other hand, as in wired local area networks (WLAN), Digital data are transmitted in their digital state.

Vertical resorption following posterior mandibular alveolar reconstruction through bone splitting and applying grafts from chin and ramus

Our present study aimed at evaluating vertical resorption following ramus and chin grafts applied in the split space after sixmonth implant delay.

Study and characterization of acoustic gain of parabolic reflector

The parabolic reflector is used in directed acoustic systems, with the aim of gettingdirectivity, as well as increasing the gain in transmission and reception. Therefore, a theoretical study was carried out to describe the factors that affect the gain of the parabolic dish in the acoustic field. The frequency-gain relationship was experimentally obtained using a parabolic dish with an elliptical aperture, in addition to the analysis of the factors that led to the deviation of the experimental relationship from the theoretical relationship.

إستراتيجية استهداف نمو عرض النقد و علاقتها بمعدل نمو الناتج القومي في سورية للفترة ما بين / 2000-2010 /

يركز البحث على تقييم استراتيجية استهداف معدل نمو عرض النقد في سورية و مدى تأثيرها و علاقتها بمعدل نمو الناتج القومي الإجمالي في الفترة ما بين أعوام 2000-2010م, و في سبيل ذلك يستعرض البحث مفهوم عرض النقد في الأدبيات الاقتصادية, و وسائل استهداف معدل نموه عبر أدوات و سياسات البنك المركزي, و مرتكزات استراتيجية استهداف عرض النقد, وصولاً إلى استعراض التجربة السورية لاستهداف معدل نمو عرض النقد ضمن الخطتين الخمسيتين خلال الفترة 2000 - 2010 م, و ذلك باستخدام الجداول و البيانات و المؤشرات المنشورة عبر البنك المركزي و المجموعة الإحصائية السورية و باستخدام البرنامج الإحصائي SPSS.