Search

Biomedical Named Entity Recognition (Bio-NER) identifies and categorises the named entities of biomedical text data such as disease, chemical, protein, and gene. Since most of the biomedical data originates from the real world, the majority of data instances do not pertain to the specific named entity of interest. Therefore, this imbalance of data adversely impacts the performance of Bio-NER using machine learning models, as their learning objective is usually dominated by the majority of non-entity tokens. Various undersampling techniques have been introduced to address this issue. Balanced Undersampling (BUS) is one of the approaches which operates at the sentence level to enhance biomedical NER (Bio-NER). However, BUS lacks in preserving contextual information during the undersampling procedure. To overcome this limitation, we introduce an improved Balanced Undersampling method (iBUS) for Bio-NER. During the undersampling process, iBUS considers the importance of individual instances and generates a balanced dataset while retaining essential instances. To validate the effectiveness of the proposed method over competitive methods, we perform experiments using the NCBI disease dataset, CHEMDNER, and BC5CDR chemical datasets. The experimental results demonstrate the superiority of the proposed method in terms of the F1 score compared to competitive approaches.

The use of encrypted data, the diversity of new protocols, and the surge in the number of malicious activities worldwide have posed new challenges for intrusion detection systems (IDS). In this scenario, existing signature-based IDS are not performing well. Various researchers have proposed machine learning-based IDS to detect unknown malicious activities based on behaviour patterns. Results have shown that machine learning-based IDS perform better than signature-based IDS (SIDS) in identifying new malicious activities in the communication network. In this paper, the authors have analyzed the IDS dataset that contains the most current common attacks and evaluated the performance of network intrusion detection systems by adopting two data resampling techniques and 10 machine learning classifiers. It has been observed that the top three IDS models—KNeighbors, XGBoost, and AdaBoost—outperform binary-class classification with 99.49%, 99.14%, and 98.75% accuracy, and XGBoost, KNneighbors, and GaussianNB outperform in multi-class classification with 99.30%, 98.88%, and 96.66% accuracy.

Deep learning is a challenging research area with respect to Text mining and Natural language processing. Deep learning is now actively used in areas such as Text classification, Text Summarization, Question Answering systems, Machine translation, Paraphrasing to name a few. The effectiveness of deep learning methods is highly influenced by hyper parameters. In this work, we illustrate the effect of Genetic Algorithm (GA) in Bidirectional Encoder Representations from Transformers (BERT) model and Robustly Optimized BERT (RoBERTa) model to optimally select the hyperparameters. The experiments are done on benchmark datasets like SMS Spam, SST-2 and Twitter airline sentiment datasets. The hyperparameter optimization experiments that we have done on these datasets have not been explored earlier. We have proposed an encoding scheme for the GA based on the parameters in consideration. We have done a comparative analysis of basic deep learning models and pretrained models based on the classification accuracy. Our work is a strong foundation for tasks like Text summarization where we can apply Evolutionary algorithms for encoding the texts so as to improve the encoding speed. From our results it is evident that using the optimal hyperparameters obtained by GA, we were able to achieve equal or above results in accuracy when compared to the baseline models. Finally, we proved using the Friedman test that the pretrained models perform better than the basic deep learning models in text classification.

The diagnosis of patients is significantly influenced by medical imaging. These images are frequently transmitted over the network for timely and accurate diagnosis. On the external networks, these medical images are prone to attack. Therefore, medical images are encrypted before transmission; this single layer of protection is not sufficient. Recently, the idea of embedding patient personal information, such as patient identity and medical records, in medical images via reversible data hiding (RDH) in the encrypted image has been proposed. Still, these schemes suffer from low embedding rates. In this paper, a RDH mechanism based on an interpolation scheme and histogram shifting is highlighted. In the proposed method, the achievable embedding capacity (EC) is 1 bit per pixel (bpp). The image encryption is achieved using blockchain technology based on 3 keys. The proposed blockchain mechanism is safe from external attacks. The results are compared with the state-of-the-art methods for both digital and medical images to validate the usefulness of the proposed method.

ABSTRACTVisceral Leishmaniasis (VL) is an insect-borne neglected disease caused by the protozoan parasite Leishmania donovani. In the absence of a commercial vaccine against VL, chemotherapy is currently the only option used for the treatment of VL. Vaccination has been considered as the most effective and powerful tool for complete eradication and control of infectious diseases. In this study, we aimed to design a peptide-based vaccine against L. donovaniusing immuno-bioinformatic tools. We identified 6 HTL, 18 CTL, and 25 B-cell epitopes from three hypothetical membrane proteins of L. donovani. All these epitopes were used to make a vaccine construct along with linkers. An adjuvant was also added at the N-terminal to enhance its immunogenicity. After that, we checked the quality of this vaccine construct and found that it is nontoxic, nonallergic, and thermally stable. A 3D structure of the vaccine construct was also generated by homology modeling to evaluate its interaction with innate immune receptors (TLR). Molecular docking was performed, which confirmed its binding with a toll-like receptor-2 (TLR-2). The stability of vaccine-TLR-2 complex and underlying interactions were evaluated using molecular dynamic simulation. Lastly, we carried out in silicocloning to check the expression of the final designed vaccine. The designed vaccine construct needs further experimental and clinical investigations to develop it as a safe and effective vaccine against VL infection.

A data warehouse system uses materialized views extensively in order to speedily tackle analytical queries. Considering that all possible views cannot be materialized due to maintenance cost and storage constraints, the selection of an appropriate set of views to materialize that achieve an optimal trade-off among query response time, maintenance cost, and the storage constraint becomes an essential necessity. The selection of such an appropriate set of views for materialization is referred to as the materialized views selection problem, which is an NP-Complete problem. In the last two decades, several new selection approaches, based on heuristics, have been proposed. Most of these have used a single objective or weighted sum approach to address the various constraints. In this article, an attempt has been made to address the bi-objective materialized view selection problem, where the objective is to minimize the view evaluation cost of materialized views and the view evaluation cost of the non-materialized views, using the Improved Strength Pareto Evolutionary Algorithm. The experimental results show that the proposed multi-objective view selection algorithm is able to select the Top-K views that achieves a reasonable trade-off between the two objectives. Materializing these selected views would reduce the query response times for analytical queries and thereby facilitates the decision-making process.

Clustering is an unsupervised classification task in data mining, which partitions data set into certain number of clusters. Though many metaheuristics have been suggested to optimize clustering results, most of them are marred by the three key issues. First, clustering performance is devised on a single validity measure; hence, it produces a single best solution biased toward the criterion; it is unacceptable as clustering has multiple conflicting criteria. Second, they find difficulty in avoiding local optima owing to lack of balancing in exploration and exploitation in the search space. Third, they require number of clusters a priori which is difficult to decide in the real-world problems. To deal with the first issue, we follow Pareto-based approach to obtain diverse trade-off solutions by optimizing conceptually contradicting validity measures sum of squared error and connectedness. The well-known swarm intelligence-based metaheuristic particle swarm optimization (PSO) has a severe drawback that it converges prematurely owing to single directional information sharing mechanism among particles in the swarm. Therefore, for the second issue, we introduce fitness predator optimizer (FPO) to enhance diversity in the PSO in multi-objective optimization scenario; it is named as FPO-MOPSO. To address the third issue, we use dynamic clustering method to decide an optimum number of clusters (k) in the range of 2 to $$\sqrt{n}$$n, where nis number of objects in the data set. We use eight real-world and two artificial data sets for comparison of the FPO-MOPSO results with three well-known competitive methods MOPSO, MABC, and NSGA-II using four quality measures convergence, diversity, coverage, and ONVG. The obtained results show superiority of the FPO-MOPSO over its competitors. Two external validity indices of clustering results F-measure and Rand index also establish the above finding. As the decision maker picks only a single solution from the set of trade-off solutions, we employ an additional measure silhouette index to select a final single solution from the archive and compare it with known number of clusters in the data sets.

In today's world, business transactional data has become the critical part of all business-related decisions. For this purpose, complex analytical queries have been run on transactional data to get the relevant information, from therein, for decision making. These complex queries consume a lot of time to execute as data is spread across multiple disparate locations. Materializing views in the data warehouse can be used to speed up processing of these complex analytical queries. Materializing all possible views is infeasible due to storage space constraint and view maintenance cost. Hence, a subset of relevant views needs to be selected for materialization that reduces the response time of analytical queries. Optimal selection of subset of views is shown to be an NP-Complete problem. In this article, a non-Pareto based genetic algorithm, is proposed, that selects Top-K views for materialization from a multidimensional lattice. An experiments-based comparison of the proposed algorithm with the most fundamental view selection algorithm, HRUA, shows that the former performs comparatively better than the latter. Thus, materializing views selected by using the proposed algorithm would improve the query response time of analytical queries and thereby facilitate in decision making.

Biomedical named entity recognition (Bio-NER) is an important and fundamental task for biomedical text mining. Bio-NER has various applications, especially in Natural Language Processing, and its performance heavily impacts dependent tasks. However, the class imbalance problem is the major issue for an effective Bio-NER. Over the past few years, rule-based methods and deep learning-based methods have been widely used. However, they are not able to effectively handle class imbalance issues. Therefore, we proposed a hybrid method by taking advantage of both methods for Bio-NER, which is named Hybrid Bio-NER (HBio-NER). Furthermore, we propose a training data format called ‘single_entity-O’, which reduces the class imbalance problem. Data preprocessing and ‘single_entity-O’ data representation format improve the class balance, further improving the HBio-NER system performance. HBio-NER identifies the named entities using the Bidirectional Long Short-Term Memory (Bi-LSTM) model and named entity boundaries (begin, inside of entity mentions) using rules. The effectiveness of the proposed method is tested on two data sets (NCBI disease and CHEMDNER chemical data sets) with different word representation models (GloVe, Word2Vec and FastText). The comparative analysis indicates that significant performance gain is achieved by the proposed method especially for disease named entity recognition.

Genetic diversity among 48 maize genotypes was determined using 11 morphological traits, 4 biochemical traits and 29 SSR primers. To assess the genetic diversity, data on morphological traits were recorded during the year 2013–2014 using standard maize descriptors. Eleven morphological traits observed across the 48 maize genotypes showed wide variation and grouped various test genotypes into two clusters. Based on the mean performance of genotypes, LM-19-07 and HKI-1348 were found superior for grain yield/plant, cob length, cob girth, kernel row/ear among all the genotypes. Protein content of LM-02-08 was the highest among all the genotypes. CML193 was found superior for iron and zinc content. Carotenoid content of LM-19-07 was higher as compared to checks. At molecular level, 29 SSR primers amplified 96 polymorphic alleles with an average of 2.16 alleles per primer. Size of amplified alleles ranged from 50 to 500 bp. Mean polymorphic information content was 0.43 showing high level of SSR polymorphism. Cluster analysis based on SSR data differentiated 48 maize genotypes into two major clusters. Bayesian model-based structure analysis assigned genotypes into two clusters and also showed the extent of admixture within individuals. Clustering pattern of maize genotypes based on SSR marker profiles was different from that of morphometric traits. Based on the pooled analysis at morphological, biochemical and molecular level, genotype LM-19-07 was found superior as well as most distant among all the accessions under study. So the genotype LM-19-07 could be used in future breeding programmes for genetic improvement in maize.

2,3-Dihydro-3β-methoxy withaferin-A (3βmWi-A) is a natural withanolide that is structurally close to withaferin-A (Wi-A), is cytotoxic to human cancer cells, and is a candidate anticancer natural compound. Using cell-based biochemical, molecular, and imaging assays, we report that Wi-A and 3βmWi-A possess contrasting activities. Whereas Wi-A caused oxidative stress to normal cells, 3βmWi-A was well tolerated at even 10-fold higher concentrations. Furthermore, it promoted survival and protected normal cells against oxidative, UV radiation, and chemical stresses. We provide molecular evidence that 3βmWi-A induces antistress and pro-survival signaling through activation of the pAkt/MAPK pathway. We demonstrate that 3βmWi-A (i) contrary to Wi-A is safe and possesses stress-relieving activity, (ii) when given subsequent to a variety of stress factors including Wi-A, protects normal cells against their toxicity, and (iii) is a vital compound that may guard normal cells against the toxicity associated with various targeted therapeutic regimes in clinical practice.

In distributed database systems, relations are replicated and fragmented at multiple sites to ensure easy availability and greater reliability. This leads to an exponential increase in the possible alternatives available for selecting the set of sites, constituting a query plan, for processing. Computing the optimal query plans, from amongst all possible query plans, is a discrete combinatorial optimization problem. This Distributed Query Plan Generation (DQPG) problem has been addressed using Bacterial Foraging Optimization (BFO) in this paper. Here, a novel BFO based DQPG algorithm (DQPGBFO), which generates the Top-K distributed query plans having the minimum total query processing cost, has been proposed. Experimental comparison of DQPGBFO with the existing Genetic Algorithm (GA) based DQPG algorithm (DQPGGA) shows that the former is able to generate Top-K query plans that have a comparatively lower total cost of processing a distributed query. This, in turn, leads to a reduction in the query response time and thus aids in decision making.

A large number of queries are posed daily against the distributed databases spread across the globe. Query processing strategies are used to generate efficient query plans for these queries. The number of such query plans increases exponentially with the increase in the number of involved sites and relations accessed by the query. Further, this complexity increases if the data is fragmented and replicated across multiple sites. This problem, referred to as the distributed query plan generation (DQPG) problem, is a combinatorial optimization problem. An attempt has been made in this paper to solve this DQPG problem using the Firefly Algorithm (FA), which is inspired by the flashing behaviour of fireflies in nature. The proposed FA based DQPG algorithm (DQPGFA), aims to generate distributed query plans incurring minimum Query Proximity Cost (QPC) value. The experimental results show that DQPGFA, in comparison to the GA based DQPG algorithm (DQPGGA), was able to select Top-K query plans that had a comparatively lesser average QPC value. Such generated query plans would, most likely, lead to an improvement in the query response time and thereby would result in effective and efficient decision making.

Prediction of high flow nasal cannula (HFNC) failure in COVID-19 patients with acute hypoxemic respiratory failure (AHRF) may improve clinical management and stratification of patients for optimal treatment. We performed a systematic review and meta-analysis to determine performance of ROX index as a predictor of HFNC failure.

"EFFI-CYCLE" derived from Efficient-Cycle promotes reducing dependency on fossil fuels for domestic transport application. The objective of this paper is to conceive, design and fabricate a three-wheel configuration (tad-pole design) vehicle powered by human-electric hybrid power and capable of seating two passengers catering to the day to day mobility needs. PTC Creo 3.0 software is used for geometric modelling of the vehicle's frame. To ensure the safety of the design the model is then imported into ANSYS V16.2 for impact analyses. Results in terms of deformation, strain and von Mises stresses are plotted. After successfully completing the design and simulation of the frame and other components it is then fabricated in the college's (S R Engineering College, Warangal) workshop.

The research carries out a parametric study to analyse slenderness effect in column design of reinforced cement concrete (RCC) flat-plate structures based on Indian standard (IS) codes namely, IS 456: 2000 and IS 875. This study is conducted on 36 columns at three locations (i.e. corner, edge and inner columns) in 12 flat-plate RCC structural models. First the methods describe in the Indian standard code for designing slender column is reviewed manually for one structural model and then FEM models are generated by using ETABS software. Parametric study is achieved by considering: four different length of column range from 3m to 6m using an increment of 1m and three slab panel's size 4.5m x 4.5m, 6m x 6m and 7.5m x 7.5m with three panels in both ways. Result shows that corner and edge column has most slenderness effect. Column height increase on and beyond 5m is substantially vulnerable for sway effect. If column length increases from one particular stage to another then steel ratio increases. In many cases, a column is not slender by considering now-sway frame but very slender by considering sway frame. For this reason, a designer should check a column for both non-sway and sway conditions.

In this article a state of the art system inversion based control (SIBC) technique is proposed for regaining the speed of a pedagogical hydrostatic drive to its desired speed under changed load torque on the drive shaft. The input calculation through the SIBC is activated only when steady state value of the output deviates from the desired level. Therefore, the desired output is attained without making the system closed loop. As a result, neither high frequency oscillation (chattering) in the control input nor the risk of closed loop instability comes into the picture as of PID controller where system output is continuously compared with the reference to calculate the input.

Railway transport dynamics cause rails in general to experience a time varying rolling contact load to add shear strain to that from traction and normal strain from the axle loads. These loading conditions together with the frictional heat at the contact interface alters the microstructure in the near-surface region of the rails and wheels. This paper presents the hardness and microstructural study of used Indian rail steel to expose the influence of load parameter active at the interface. Samples are prepared from the rail section taken from Dhanbad railway division. The hardness variation along the rail head cross-section and metallurgical variations of the test sample are presented. Microstructural examinations reveal existence of micro cracks near to the rail surface favouring formation of wear debris. Also, the presence of the pro-eutectoid ferrite and cementite phase at the grain boundaries of the pearlitic rail material is observed. The fatigue cracks are found to initiate across the grain boundaries containing the pro-eutectoid ferrite phase. This study will help to develop a better understanding of rolling contact fatigue (RCF) crack initiation and wear debris formation mechanisms on the rails. Thus, these results can be used to improve rail maintenance planning and risk assessments.