Distribution Of Different Classes Research Articles

Pesticides in surface water of the Ondas river watershed, western Bahia, Brazil: Spatial-seasonal distribution and risk assessment

This study investigated the occurrence and seasonal distribution of different classes of pesticides in surface waters of the Ondas River Watershed, as well as potential risks to the aquatic health and human water consumption in the western region of Bahia state, Brazil. Two gas chromatography-mass spectrometry analytical methods were applied to monitor 34 pesticides in water samples collected during both the dry and rainy seasons at 17 sites. Upon individual analysis, only γ-HCH, methoxychlor, demeton-S, methyl parathion, fenitrothion, chlorpyrifos, and azoxystrobin exhibited statistically significant differences between seasons. During rainy season, concentration medians of residues were higher for γ-HCH (74.7 ng L−1), methoxychlor (25.1 ng L−1), and azoxystrobin (47.2 ng L−1), potentially linked to historical contamination or illegal use. Conversely, pesticides like methyl parathion, fenitrothion, and chlorpyrifos, belonging to the organophosphate class, showed higher concentration medians in the dry period, measuring 75.1, 5.50, and 10.8 ng L−1, respectively, probably due to region crop activities. The risk quotient (RQ) assessment for aquatic life indicated that 59.0% of the samples in the dry season and 76.0% in the rainy season had RQ values greater than one, signifying a critical scenario for species conservation. Regarding human consumption, elevated risks were observed for heptachlor in both sampling periods and for azoxystrobin during the rainy season, surpassing RQ levels above 1, indicating danger in untreated water ingestion. Additionally, 24.0% and 53.0% of the samples in the dry and rainy seasons, respectively, contained at least one pesticide exceeding the EU resolution limit (100 ng L−1). Therefore, considering this information, implementing mitigation measures to avoid the river's contamination becomes imperative.

Ecological significance of Candidatus ARS69 and Gemmatimonadota in the Arctic glacier foreland ecosystems.

The Gemmatimonadota phylum has been widely detected in diverse natural environments, yet their specific ecological roles in many habitats remain poorly investigated. Similarly, the Candidatus ARS69 phylum has been identified only in a few habitats, and literature on their metabolic functions is relatively scarce. In the present study, we investigated the ecological significance of phyla Ca. ARS69 and Gemmatimonadota in the Arctic glacier foreland (GF) ecosystems through genome-resolved metagenomics. We have reconstructed the first high-quality metagenome-assembled genome (MAG) belonging to Ca. ARS69 and 12 other MAGs belonging to phylum Gemmatimonadota from the three different Arctic GF samples. We further elucidated these two groups phylogenetic lineage and their metabolic function through phylogenomic and pangenomic analysis. The analysis showed that all the reconstructed MAGs potentially belonged to novel species. The MAGs belonged to Ca. ARS69 consist about 8296 gene clusters, of which only about 8% of single-copy core genes (n = 980) were shared among them. The study also revealed the potential ecological role of Ca. ARS69 is associated with carbon fixation, denitrification, sulfite oxidation, and reduction biochemical processes in the GF ecosystems. Similarly, the study demonstrates the widespread distribution of different classes of Gemmatimonadota across wide ranges of ecosystems and their metabolic functions, including in the polar region. KEY POINTS: • Glacier foreland ecosystems act as a natural laboratory to study microbial community structure. • We have reconstructed 13 metagenome-assembled genomes from the soil samples. • All the reconstructed MAGs belonged to novel species with different metabolic processes. • Ca. ARS69 and Gemmatimonadota MAGs were found to participate in carbon fixation and denitrification processes.

Employing synthetic data for addressing the class imbalance in aspect-based sentiment classification

ABSTRACT The class imbalance problem, in which the distribution of different classes in training data is unequal or skewed, is a prevailing problem. This can lead to classifier algorithms being biased, negatively impacting the performance of the minority class. In this paper, we addressed the class imbalance problem in datasets for aspect-based sentiment classification. Aspect-based Sentiment Classification (AbSC) is a type of fine-grained sentiment analysis in which sentiments about particular aspects of an entity are extracted. In this work, we addressed the issue of class imbalance by creating synthetic data. For synthetic data generation, two techniques have been proposed: paraphrasing using the PEGASUS fine-tuned model and backtranslation using the M2M100 neural machine translation model. We compared these techniques with two other class balancing techniques, such as weighted oversampling and cross-entropy loss with class weight. An extensive experimental study has been conducted on three benchmark datasets for restaurant reviews: SemEval-2014, SemEval-2015, and SemEval-2016. We applied these methods to the BERT-based deep learning model for aspect-based sentiment classification and studied the effect of balancing the data on the performance of these models. Our proposed balancing technique, using synthetic data, yielded better results than the other two existing methods for dealing with multi-class imbalance.

PCCT: Progressive Class-Center Triplet Loss for Imbalanced Medical Image Classification.

Imbalanced training data in medical image diagnosis is a significant challenge for diagnosing rare diseases. For this purpose, we propose a novel two-stage Progressive Class-Center Triplet (PCCT) framework to overcome the class imbalance issue. In the first stage, PCCT designs a class-balanced triplet loss to coarsely separate distributions of different classes. Triplets are sampled equally for each class at each training iteration, which alleviates the imbalanced data issue and lays solid foundation for the successive stage. In the second stage, PCCT further designs a class-center involved triplet strategy to enable a more compact distribution for each class. The positive and negative samples in each triplet are replaced by their corresponding class centers, which prompts compact class representations and benefits training stability. The idea of class-center involved loss can be extended to the pair-wise ranking loss and the quadruplet loss, which demonstrates the generalization of the proposed framework. Extensive experiments support that the PCCT framework works effectively for medical image classification with imbalanced training images. On four challenging class-imbalanced datasets (two skin datasets Skin7 and Skin 198, one chest X-ray dataset ChestXray-COVID, and one eye dataset Kaggle EyePACs), the proposed approach respectively obtains the mean F1 score 86.20, 65.20, 91.32, and 87.18 over all classes and 81.40, 63.87, 82.62, and 79.09 for rare classes, achieving state-of-the-art performance and outperforming the widely used methods for the class imbalance issue.

Estimating Multiclass Service Demand Distributions Using Markovian Arrival Processes

Building performance models for software services in DevOps is costly and error-prone. Accurate service demand distribution estimation is critical to precisely modeling queueing behaviors and performance prediction. However, current estimation methods focus on capturing the mean service demand, disregarding higher-order moments of the distribution that still can largely affect prediction accuracy. To address this limitation, we propose to estimate higher moments of the service demand distribution for a microservice from monitoring traces. We first generate a closed queueing model to abstract software performance and use it to model the departure process of requests completed by the software service as a Markovian arrival process (MAP). This allows formulating the estimation of service demand into an optimization problem, which aims to find the first multiple moments of the service demand distribution that maximize the likelihood of the MAP using generated the measured inter-departure times. We then estimate the service demand distribution for different classes of service with a maximum likelihood algorithm and novel heuristics to mitigate the computational cost of the optimization process for scalability. We apply our method to real traces from a microservice-based application and demonstrate that its estimations lead to greater prediction accuracy than exponential distributions assumed in traditional service demand estimation approaches for software services.

The linkage between prevalence of integron I and reduced susceptibility to biocides in MDR Klebsiella pneumoniae isolated from neonates.

Klebsiella pneumoniae causes challenging nosocomial fatal infections including neonatal sepsis. Our study aims at clarifying the contribution of integrons in the observed reduced susceptibility of multidrug-resistant (MDR) K. pneumoniae isolated from septicemic neonates to the clinically used antimicrobial agents and biocides. Eighty-six K. pneumoniae isolates were collected from Mansoura University Children's Hospital from septicemic neonates. Isolates were subjected to antibiotic and biocide susceptibility using disk diffusion and the agar dilution method, respectively. The distribution of different classes of integrons was screened in the isolates by PCR. Detected inegron I was sequenced in selected isolates. Fifty-seven isolates (66.27%) were MDR. In the MDR isolates, class I integron was detected in 23 (40.3%), integron III was detected in 20 (35%), whereas integron II could not be detected. Sequencing results of integron I from MDR K. pneumoniae isolates revealed that only aminoglycoside and folate synthesis inhibitors gene cassettes were detected, while the rest of the resistance genes were not associated with integron I. The presence of integron I in MDR K. pneumoniae tested isolates may contribute only to some biocide resistance, however, it does not seem to be the only contributor in multiple drug resistance.

Tissue-preferential recruitment of electron transfer chains for cytochrome P450-catalyzed phenolic biosynthesis.

Cytochrome P450 system consists of P450 monooxygenase and redox pattern(s). While the importance of monooxygenases in plant metabolism is well documented, the metabolic roles of the related redox components have been largely overlooked. Here, we show that distinct electron transfer chains are recruited in phenylpropanoid-monolignol P450 systems to support the synthesis and distribution of different classes of phenolics in different plant tissues. While Arabidopsis cinnamate 4-hydroxylase adopts conventional NADPH-cytochrome P450 oxidoreductase (CPR) electron transfer chain for its para-hydroxylation reaction, ferulate 5-hydroxylase uses both NADPH-CPR-cytochrome b5 (CB5) and NADH-cytochrome b5 reductase-CB5 chains to support benzene ring 5-hydroxylation, in which the former route is primarily recruited in the stem for syringyl lignin synthesis, while the latter dominates in the syntheses of 5-hydroxylated phenolics in seeds and seed coat suberin. Our study unveils an additional layer of complexity and versatility of P450 system that the plants evolved for diversifying phenolic repertoires.

Polish Comparative Adjectives and Adverbs

The paper investigates the morphosyntax of Polish synthetic comparative adjectives and adverbs. It is argued that we can predict the distribution of different classes of adjectival roots and suffixes if we adopt the idea that both types of morphemes lexicalize syntactic constituents, the central tenet of Nanosyntax. The paper makes a case for two central claims. One is that the syn-sem properties of adjectives can be described with a finegrained syntactic sequence proposed for Slovak in Vanden Wyngaerd et al. (2020). The other one is that the lexical properties of Polish gradable adverbs follow from the syntactic representation of the adverb as properly containing the syntactic representation of the adjective.

A multi-band centroid contrastive reconstruction fusion network for motor imagery electroencephalogram signal decoding.

Motor imagery (MI) brain-computer interface (BCI) assist users in establishing direct communication between their brain and external devices by decoding the movement intention of human electroencephalogram (EEG) signals. However, cerebral cortical potentials are highly rhythmic and sub-band features, different experimental situations and subjects have different categories of semantic information in specific sample target spaces. Feature fusion can lead to more discriminative features, but simple fusion of features from different embedding spaces leading to the model global loss is not easily convergent and ignores the complementarity of features. Considering the similarity and category contribution of different sub-band features, we propose a multi-band centroid contrastive reconstruction fusion network (MB-CCRF). We obtain multi-band spatio-temporal features by frequency division, preserving the task-related rhythmic features of different EEG signals; use a multi-stream cross-layer connected convolutional network to perform a deep feature representation for each sub-band separately; propose a centroid contrastive reconstruction fusion module, which maps different sub-band and category features into the same shared embedding space by comparing with category prototypes, reconstructing the feature semantic structure to ensure that the global loss of the fused features converges more easily. Finally, we use a learning mechanism to model the similarity between channel features and use it as the weight of fused sub-band features, thus enhancing the more discriminative features, suppressing the useless features. The experimental accuracy is 79.96% in the BCI competition Ⅳ-Ⅱa dataset. Moreover, the classification effect of sub-band features of different subjects is verified by comparison tests, the category propensity of different sub-band features is verified by confusion matrix tests and the distribution in different classes of each sub-band feature and fused feature are showed by visual analysis, revealing the importance of different sub-band features for the EEG-based MI classification task.

DEEP REINFORCEMENT LEARNING WITH HIDDEN MARKOV MODEL FOR SPEECH RECOGNITION

Nowadays, many applications uses speech recognition especially the field of computer science and electronics, Speech Recognition (SR) is the interpretation of words spoken into a text. It is also known as Speech-To-Text (STT) or Automatic-Speech-Recognition(ASR), or just Word-Recognition(WR). The Hidden-Markov-Model (HMM) is a type of Markov model, which means that the future state of the model depends on the current state, not on the entire history of the system and the goal of HMM is to learn a sequence of hidden states from a set of known states. The Long-Short-Time-Memory (LSTM) network is a type of Recurrent Neural Network (RNN) that can learn long-term dependencies between time steps of sequence data. The LSTM network is trained by the network in order to predict the values of subsequent time steps in a series-to-series regression. Deep Neural Network (DNN) models are better classifiers than Gaussian Mixture Models (GMMs), they can generalize much better with a smaller number of parameters over complex distributions. They model distributions of different classes jointly, called “distributed” learning, or, more properly “tied” learning. This work is aimed at developing a speech recognition model that will predict isolated speech of some selected fruits in Hausa, Igbo and Yoruba language by using the predicting power of Mel-Frequency-Cepstral-Coefficient (MFCC), LSTM and HMM algorithms. The findings of the study would improve the development of better automatic speech applications systems and would benefit the academic and research community in the field of Natural Language Processing.

Adversarial training with distribution normalization and margin balance

Adversarial training is the most effective method to improve adversarial robustness. However, it does not explicitly regularize the feature space during training. Adversarial attacks usually move a sample iteratively along the direction which causes the steepest ascent of classification loss by crossing decision boundary. To alleviate this problem, we propose to regularize the distributions of different classes to increase the difficulty of finding an attacking direction. Specifically, we propose two strategies named Distribution Normalization (DN) and Margin Balance (MB) for adversarial training. The purpose of DN is to normalize the features of each class to have identical variance in every direction, in order to eliminate easy-to-attack intra-class directions. The purpose of MB is to balance the margins between different classes, making it harder to find confusing class directions (i.e., those with smaller margins) to attack. When integrated with adversarial training, our method can significantly improve adversarial robustness. Extensive experiments under white-box, black-box, and adaptive attacks demonstrate the effectiveness of our method over other state-of-the-art methods.

Hellenistic and Roman Gerasa: The Archaeology and History of a Decapolis City. Edited by Achim Lichtenberger and Rubina Raja. Turnhout: Brepols 2020. Pp. xviii + 390. €110. ISBN 978-2-503-58504-8 (paper).

<i>Hellenistic and Roman Gerasa: The Archaeology and History of a Decapolis City</i>. Edited by Achim Lichtenberger and Rubina Raja. Turnhout: Brepols 2020. Pp. xviii + 390. €110. ISBN 978-2-503-58504-8 (paper).

Filtering Properties of Discrete and Continuous Elastic Systems in Series and Parallel

Filtering properties and local energy distribution in different classes of periodic micro-structured elastic systems are analysed in this work. Out-of-plane wave propagation is considered in continuous and discrete elastic systems arranged in series and parallel. Filtering properties are determined from the analysis of dispersion diagrams and energy distribution within different phases in the representative unit cell. These are determined analytically by implementing a transfer matrix formalism. The analysis given in the work indicates quantitatively how to couple phases, having discrete and continuous nature, in order to tune wave propagation and energy localisation.

On IoT intrusion detection based on data augmentation for enhancing learning on unbalanced samples

Internet of things (IoT) security is a prerequisite for the rapid development of the IoT to enhance human well-being. Machine learning-based intrusion detection systems (IDS) have good protection capabilities. However, it is difficult to identify attack information in massive amounts of data, which leads to inefficient model detection when faced with insufficient samples for certain types of attacks. In this regard, this paper fuses deep learning methods and statistical ideas to address the problem of minority samples attack detection, and proposes an intrusion detection method for the IoT based on Improved Conditional Variational Autoencoder (ICVAE) and Borderline Synthetic Minority Oversampling Technique (BSM), called ICVAE-BSM. By introducing an auxiliary network into the Conditional Variational Autoencoder (CVAE) to adjust the output probability distribution of the encoder, learning the posterior distribution of different classes of samples, so that the distributions of samples of the same class are concentrated, and the distributions of different classes of samples are scattered in the latent space; then based on BSM, adaptively synthesize the edge latent variables in the latent space of ICVAE, and feed the new synthetic edge latent variables to the ICVAE’s decoder to generate representative new samples to balance the data set. The output of the encoder is connected to the Softmax classifier at last, and the original samples are mixed with the generated samples to fine-tune it to enhance its generalization ability for intrusion detection of minority samples. We use the NSL-KDD data set, CIC-IDS2017 data set and CSE-CIC-IDS2018 data set to simulate and evaluate the model, the experimental results show that the proposed method can more effectively improve the accuracy of IoT attack detection under the condition of unbalanced samples.

Unveiling the genomic potential of Pseudomonas type strains for discovering new natural products.

Microbes host a huge variety of biosynthetic gene clusters that produce an immeasurable array of secondary metabolites with many different biological activities such as antimicrobial, anticarcinogenic and antiviral. Despite the complex task of isolating and characterizing novel natural products, microbial genomic strategies can be useful for carrying out these types of studies. However, although genomic-based research on secondary metabolism is on the increase, there is still a lack of reports focusing specifically on the genus Pseudomonas . In this work, we aimed (i) to unveil the main biosynthetic systems related to secondary metabolism in Pseudomonas type strains, (ii) to study the evolutionary processes that drive the diversification of their coding regions and (iii) to select Pseudomonas strains showing promising results in the search for useful natural products. We performed a comparative genomic study on 194 Pseudomonas species, paying special attention to the evolution and distribution of different classes of biosynthetic gene clusters and the coding features of antimicrobial peptides. Using EvoMining, a bioinformatic approach for studying evolutionary processes related to secondary metabolism, we sought to decipher the protein expansion of enzymes related to the lipid metabolism, which may have evolved toward the biosynthesis of novel secondary metabolites in Pseudomonas . The types of metabolites encoded in Pseudomonas type strains were predominantly non-ribosomal peptide synthetases, bacteriocins, N-acetylglutaminylglutamine amides and ß-lactones. Also, the evolution of genes related to secondary metabolites was found to coincide with Pseudomonas species diversification. Interestingly, only a few Pseudomonas species encode polyketide synthases, which are related to the lipid metabolism broadly distributed among bacteria. Thus, our EvoMining-based search may help to discover new types of secondary metabolite gene clusters in which lipid-related enzymes are involved. This work provides information about uncharacterized metabolites produced by Pseudomonas type strains, whose gene clusters have evolved in a species-specific way. Our results provide novel insight into the secondary metabolism of Pseudomonas and will serve as a basis for the prioritization of the isolated strains. This article contains data hosted by Microreact.

Insight into Potential Anticancer Activity of Algal Flavonoids: Current Status and Challenges.

Flavonoids are some of the most precious phytochemicals, believed to be found largely in terrestrial plants. With the advancement of phytochemical research and marine bioprospecting, flavonoids have also been reported by the research of microalgae and macroalgae. High growth rate with minimal nutritional and growth requirement, saving arable land and rich metabolic profile make microalgae an excellent repertoire of novel anticancer compounds, such as flavonoids. In addition, marine algae, especially seaweeds contain different types of flavonoids which are assumed to have unique chemical structures and bioactivities than their terrestrial counterparts. Flavonoids are not only good antioxidants but also have the abilities to kill cancer cells by inducing apoptosis and autophagy. However, the study of the anticancer properties of flavonoids is largely limited to terrestrial plants. This review offers an insight into the distribution of different classes of flavonoids in eukaryotic microalgae, cyanobacteria and seaweeds with their possible anticancer activities. In addition, extraction and purification methods of these flavonoids have been highlighted. Finally, prospects and challenges to use algal flavonoids as anticancer agents have been discussed.

THE STUDY OF QUALITY INDICATORS AND FRACTIONAL COMPOSITION OF WHEAT GRAIN PROTEIN OF SOUTHERN REGIONS OF UKRAINE

The results of studies of the quality of food and non-food grain of wheat from the southern regions of Ukraine in 2016 harvest are presented. The indicators of the quality of wheat grain were determined, the uniformity of their distribution in different classes of wheat was assessed, the content of protein fractions of the gluten complex and the molecular weights of individual protein structures were determined. It was found that the studied 13 samples of wheat grain had the following quality indicators: moisture content 9.4 ... 13.5%, bulk density 731 ... 814 g/l, protein content 10.1 ... 13.2%, gluten content 16.4 ... 25.6%, gluten quality 50 ... 110 units of the IDK device, falling number 311 ... 493 s, grain admixture 1.2 ... 10.5%, grains damaged by a wheat bug, 0.1 ... 10.1%. It was noted that the bulk density of all samples of wheat of 2 ... 6 classes exceeded the standard values in each class. In terms of protein content, 2 from 3 samples of class 5 wheat and 1 from 2 samples of 6 class exceeded the standard value of class 3 wheat (≥11.0%). The same was observed for these samples and for the gluten content (≥18%). Almost all grain samples, except for 2 samples of the 5th class and 2 samples of the 6th class, met the requirements of wheat of the 2nd class (45 ... 100 units device VDK). The falling number of all samples was high and significantly exceeded the normalized value (180 s), even for class 2 wheat. It was found that the uniformity of the distribution of individual quality indicators in different classes of wheat grain by the coefficient of variation is not the same. Such indicators of grain quality as moisture content, bulk density and protein content in different samples of the south of Ukraine in 2016 harvest fluctuated weakly (V≤10%), in terms of the amount of gluten and the falling number fluctuations were average (10% ≤V≤20%). Fluctuations in the content of grain impurities and grains damaged by the wheat bug are big (V≥20%). For the last indicator, fluctuations relative to the average are very large and exceed 120 %. This indicates that in different areas where wheat was grown, the infestation by the wheat bug was very different. It was shown that regardless of the total amount of protein determined by the Infratec FOSS express analyzer, four protein fractions of the gluten complex (albumins, globulins, gliadins and glutenins) are present in all grain classes, and the molecular weights of the protein structures of the main fractions are within small limits, which are in kDa - for albumins and globulins 20...30, gliadins 27.. 100 and glutenins 30...35. It was noted that in the varietal wheat "Chernobrova" albumins and globulins have a slightly higher molecular weight (30...40 kDa) than in non-varietal wheat.

Emergence of Carbapenemase Encoding Genes in Proteus Species in Tertiary Care Hospital of Bangladesh

Along with different carbapenemase encoding genes, in recent year class D OXA enzymes are documented in Proteus spp which are not common in Enterobacteriaceae. The dissemination of plasmids, transposons and integrons among bacteria and species playing roles for this dissemination. So, this study was designed to observe the emergence and distribution of different classes of carbapenemase encoding genes among imipenem resistant Proteus spp. isolated from tertiary care hospital in Bangladesh. Total 15 imipenem resistant Proteus isolates were included in this study, which were collected from wound swab, pus, urine and blood samples. Identification was done by culture and biochemical test and antibiotic susceptibility test was done by disc diffusion method. MIC of imipenem (g/ml) was done among imipenem resistant P. mirabilis by agar dilution method. blaKPC, blaNDM-1, blaVIM, blaIMP, blaOXA-48 like, blaOXA-23 like, blaOXA-51 like, blaOXA-58 like carbapenemase encoding genes were detected among imipenem resistant Proteus spp. by PCR and sequencing of blaOXA-48 like, blaOXA-51 like gene done by capillary method to compare the sequences with the same gene, available in gene bank. Among 15 imipenem resistant isolates blaNDM-1 (26.67%), blaKPC (20%), blaVIM (20%), blaOXA-484 (20%) were predominant carbapenemase encoding genes followed by blaOXA-66(6.67%). This study finds that blaOXA-484 gene and blaOXA-66 class D carbapenemase encoding genes are emerging in Proteus spp. and may play a contributing factor in developing carbapenem resistance. Bangladesh Med J. 2020 May; 49(2) : 1-8

Leveraging the Bhattacharyya coefficient for uncertainty quantification in deep neural networks

Modern deep learning models achieve state-of-the-art results for many tasks in computer vision, such as image classification and segmentation. However, its adoption into high-risk applications, e.g. automated medical diagnosis systems, happens at a slow pace. One of the main reasons for this is that regular neural networks do not capture uncertainty. To assess uncertainty in classification, several techniques have been proposed casting neural network approaches in a Bayesian setting. Amongst these techniques, Monte Carlo dropout is by far the most popular. This particular technique estimates the moments of the output distribution through sampling with different dropout masks. The output uncertainty of a neural network is then approximated as the sample variance. In this paper, we highlight the limitations of such a variance-based uncertainty metric and propose an novel approach. Our approach is based on the overlap between output distributions of different classes. We show that our technique leads to a better approximation of the inter-class output confusion. We illustrate the advantages of our method using benchmark datasets. In addition, we apply our metric to skin lesion classification—a real-world use case—and show that this yields promising results.

Adaptive Selecting and Learning Network and a New Benchmark for Imbalanced Fine-Grained Ship Classification

Ship classification is one of the most essential tasks in ship surveillance, which is an important but challenging problem. Most existing methods are designed for remote sensing images and there are few works processing natural ship images captured by camera. In this paper, we design a new framework called Adaptive Selecting and Learning Network (ASL), to solve the problem of fine-grained classification of ship natural images in the real world. Our method has two key contributions to enable more effective learning from the imbalanced ship data. First, we present a memory network equipped with an adaptive selecting learning strategy to selectively memorize the hard samples that are difficult to classify. The presented learning strategy can re-balance the data distribution of different classes in the training procedure, which achieves more effective learning. Second, we design an inference network with an attention mechanism, to capture the structural similarities between new samples and hard samples. The attention mechanism can enhance the training of hard samples and yields better learning performance. Moreover, our framework can be easily combined with existing fine-grained methods. In addition, we propose a new Dachan Island Ship (DIS) dataset, which is collected in the real-world scenarios. The DIS dataset has a significant imbalanced distribution between classes that aligns with the real situation. Extensive experimental results on the proposed DIS show that our model outperforms most of the existing fine-grained classification methods.