Data scarcity is a major challenge when training deep learning (DL) models. DL demands a large amount of data to achieve exceptional performance. Unfortunately, many applications have small or inadequate data to train DL frameworks. Usually, manual labeling is needed to provide labeled data, which typically involves human annotators with a vast background of knowledge. This annotation process is costly, time-consuming, and error-prone. Usually, every DL framework is fed by a significant amount of labeled data to automatically learn representations. Ultimately, a larger amount of data would generate a better DL model and its performance is also application dependent. This issue is the main barrier for many applications dismissing the use of DL. Having sufficient data is the first step toward any successful and trustworthy DL application. This paper presents a holistic survey on state-of-the-art techniques to deal with training DL models to overcome three challenges including small, imbalanced datasets, and lack of generalization. This survey starts by listing the learning techniques. Next, the types of DL architectures are introduced. After that, state-of-the-art solutions to address the issue of lack of training data are listed, such as Transfer Learning (TL), Self-Supervised Learning (SSL), Generative Adversarial Networks (GANs), Model Architecture (MA), Physics-Informed Neural Network (PINN), and Deep Synthetic Minority Oversampling Technique (DeepSMOTE). Then, these solutions were followed by some related tips about data acquisition needed prior to training purposes, as well as recommendations for ensuring the trustworthiness of the training dataset. The survey ends with a list of applications that suffer from data scarcity, several alternatives are proposed in order to generate more data in each application including Electromagnetic Imaging (EMI), Civil Structural Health Monitoring, Medical imaging, Meteorology, Wireless Communications, Fluid Mechanics, Microelectromechanical system, and Cybersecurity. To the best of the authors’ knowledge, this is the first review that offers a comprehensive overview on strategies to tackle data scarcity in DL.
Multilayer reservoirs are currently modeled as a single zone system by averaging the reservoir parameters associated with each reservoir zone. However, this type of modeling is rarely accurate because a single zone system does not account for the fact that each zone's pressure decreases independently. Pressure drop for each zone has an effect on the total output and would result in inter-flow and the premature depletion of one of the zones. Understanding reservoir performance requires a precise estimation of each layer's permeability and skin factor. The Multilayer Transient Analysis is a well-testing technique designed to determine formation properties in more than one layer, and its effectiveness over the past two decades has been
... Show MoreA new modified differential evolution algorithm DE-BEA, is proposed to improve the reliability of the standard DE/current-to-rand/1/bin by implementing a new mutation scheme inspired by the bacterial evolutionary algorithm (BEA). The crossover and the selection schemes of the DE method are also modified to fit the new DE-BEA mechanism. The new scheme diversifies the population by applying to all the individuals a segment based scheme that generates multiple copies (clones) from each individual one-by-one and applies the BEA segment-wise mechanism. These new steps are embedded in the DE/current-to-rand/bin scheme. The performance of the new algorithm has been compared with several DE variants over eighteen benchmark functions including sever
... Show MoreThe present study was conducted on spring-fed(Abdullah Abu-Nadjem Oasis)west of AlDewanyah province,Iraq. ,during Winter 2007. The results indicate that water quality was neutral tend to be slightly alkaline pH was 7.7 and 7.4 for station(1)and station (2) respectively. Total hardness value of water about 1520 and 1488 mg/l.for two stations respectively ,This high values of hardness may be due to Calcium ions 397.5 and 395.5mg/l.for two stations. According to Salinity values (1.58 - 2.22) ‰(ppt) ,water of oasis may be classified as brackish water, the oasis water was well aerated dissolved oxygen was 5.3 and 7.4 mg/l. On the other hand about (58) algal taxa were recorded
... Show MoreSemiconductor-based metal oxide gas detector of five mixed from zinc chloride Z and tin chloride S salts Z:S ratio 0, 25, 50, 75 and 100% were fabricated on glass substrate by a spray pyrolysis technique. With thickness were about 0.2 ±0.05 μm using water soluble as precursors at a glass substrate temperature 500 ºC±5, 0.05 M, and their gas sensing properties toward CH4, LPG and H2S gas at different concentration (10, 100, 1000 ppm) in air were investigated at room temperature which related with the petroleum refining industry.
Furthermore structural and morphology properties were scrutinize. Results shows that the mixing ratio affect the composition of formative oxides were (ZnO, Zn2SnO4, Zn2SnO4+ZnSnO3, ZnSnO3, SnO2) ratios ment
Carbon nanoparticles are prepared by sonication using carbon black powder. The surface morphology of carbon black (CB) and carbon nanoparticles (CNPs) is investigated using scanning electron microscopy (SEM). The particles size ranges from 100 nm to 400 nm for CB and from 10 nm to 100 nm for CNPs. CNPs and CB are mixed with silicon glue of different ratios of 0.025, 0.2, 0.05, and 0.1 to synthesis films. The optical properties of the prepared films are investigated through reflectance and absorbance analyses. The ratio of 0.05 for CNPs and CB is the best for solar paint because of its higher solar water heater efficiency and is then added to the silicon glue . Temperature of cold water and temperature of hot water in storage tank were ta
... Show MoreThe advent of UNHCR reports has given rise to the uniqueness of its distinctive way of image representation and using semiotic features. So, there are a lot of researches that have investigated UNHCR reports, but no research has examined images in UNHCR reports of displaced Iraqis from a multimodal discourse perspective. The present study suggests that the images are, like language, rich in many potential meanings and are governed by clearly visual grammar structures that can be employed to decode these multiple meanings. Seven images are examined in terms of their representational, interactional and compositional aspects. Depending on the results, this study concludes that the findings support the visual grammar theory and highlight the va
... Show MoreHomomorphic encryption became popular and powerful cryptographic primitive for various cloud computing applications. In the recent decades several developments has been made. Few schemes based on coding theory have been proposed but none of them support unlimited operations with security. We propose a modified Reed-Muller Code based symmetric key fully homomorphic encryption to improve its security by using message expansion technique. Message expansion with prepended random fixed length string provides one-to-many mapping between message and codeword, thus one-to many mapping between plaintext and ciphertext. The proposed scheme supports both (MOD 2) additive and multiplication operations unlimitedly. We make an effort to prove
... Show More