Explore the vast range of titles available.

Discover the concepts of deep learning used for natural language processing (NLP), with full-fledged examples of neural network models such as recurrent neural networks, long short-term memory networks, and sequence-2-sequence models. You'll start by covering the mathematical prerequisites and the fundamentals of deep learning and NLP with practical examples. The first three chapters of the book cover the basics of NLP, starting with word-vector representation before moving onto advanced algorithms. The final chapters focus entirely on implementation, and deal with sophisticated architectures such as RNN, LSTM, and Seq2seq, using Python tools: TensorFlow, and Keras. Deep Learning for Natural Language Processing follows a progressive approach and combines all the knowledge you have gained to build a question-answer chatbot system. This book is a good starting point for people who want to get started in deep learning for NLP. All the code presented in the book will be available in the form of IPython notebooks and scripts, which allow you to try out the examples and extend them in interesting ways. You will: Gain the fundamentals of deep learning and its mathematical prerequisites Discover deep learning frameworks in Python Develop a chatbot Implement a research paper on sentiment classification.

Every year, Coronary Artery Disease (CAD) claims lives of over a million people. CAD occurs when the coronary arteries, responsible for supplying oxygenated blood to the heart, get occluded due to plaque deposits on their inner walls. The most critical fact about this disease is that it develops gradually over the years and by the time symptomatic changes such as angina or shortness of breath appear, the disease has already become severe. The overall aim of the proposed work is to detect CAD efficiently in its early stage while utilizing (radial) arterial blood pressure (ABP) along with photoplethysmogram (PPG) signals so that necessary clinical measures may be taken timely. To achieve this objective, firstly, ABP and PPG data of 73 CAD and 64 non-CAD (not suffering from any cardiac condition) subjects have been collected from MIMIC-II waveform database with matched subset. Secondly, the collected data is pre-processed using band pass filters having bandwidths of 2.5 to 16 Hz and 1.5 to 16 Hz for ABP and PPG respectively. Thirdly, nineteen features have been extracted from each of the two signals; some of the key features include mean of pulse duration, mean of rising slope and ratio of low frequency to high frequency. Finally, extensive analysis on CAD and non-CAD classification is carried out on the basis of extracted features while employing state-of-the-art classifiers such as support vector machines (SVM), K-nearest neighbors (KNN) and neural networks(NN). The numerical experiments have led to the interpretation that neural network outperforms other classifiers, claiming an accuracy of about 90%. Moreover, accuracy of the proposed approach is found to be better than the state-of-the-art works reported in literature where one of or combinations of cardiovascular signals, namely, electrocardiogram (ECG), phonocardiogram (PCG) and photoplethysmogram (PPG) have been utilized for the CAD detection.

This research highlights the importance of the prefrontal theta-EEG rhythm in sustained attention monitoring over the Fp1 electrode. In an experiment conducted with 20 participants, four successive mental tasks are sent briefly by an automated computer program connected to a speakerphone: wait, relax, get ready, and concentrate. Furthermore, each individual participated in this experiment 20 times. The result is determined by how well the individual performed on the task and by examining the collected data. Subjects who start to focus on a target in fewer than 100 s are considered high-focused, and those who take more than 100 s are referred to as low-focused. The gamma, beta, alpha, and theta EEG rhythms are classified using multi-stage discrete wavelet transform for the high-focused and low-focused subjects. Then, eight statistical features are computed for the theta, alpha, beta, and gamma rhythms for the high-focused and low-focused subjects. Finally, these features train the proposed model with a 55% training and 45% testing ratio. The K-Nearest Neighbour (KNN), a machine learning classifier, is applied to classify these features. The research findings are (a) that the KNN classifier attained the best f1-score of 88.88% for theta-EEG rhythm, (b) additionally, the KNN classifier got 85.71% f1-score with alpha-EEG rhythm, 66.66% f1-score with beta, and gamma EEG rhythms, and 53.33% f1-score with the combination of all the EEG rhythms (theta, alpha, beta, and gamma). This research concludes that the theta-EEG rhythm is highly relevant in identifying the human “attentive state” compared to other EEG rhythms.

[This corrects the article DOI: 10.1007/s11571-024-10113-0.].

Livestock plays an essential role in sustaining human livelihoods, offering a diverse range of species integral to food security, economic stability, and cultural traditions. The domestication of livestock, which began over 10,000 years ago, has driven significant genetic changes in species such as cattle, buffaloes, sheep, goats, and pigs. Recent advancements in genomic technologies, including next-generation sequencing (NGS), genome-wide association studies (GWAS), and genomic selection, have dramatically enhanced our understanding of these genetic developments. This review brings together key research on the domestication process, phylogenetics, genetic diversity, and selection signatures within major livestock species. It emphasizes the importance of admixture studies and evolutionary forces like natural selection, genetic drift, and gene flow in shaping livestock populations. Additionally, the integration of machine learning with genomic data offers new perspectives on the functional roles of genes in adaptation and evolution. By exploring these genomic advancements, this review provides insights into genetic variation and evolutionary processes that could inform future approaches to improving livestock management and adaptation to environmental challenges, including climate change.

Freshly excavated overburden (spoils) dumped during open-cut coal mining generate saline leachate that can lead to environmental impacts. Predictions of leachate salinity remain uncertain, largely due to incomplete knowledge of responses of spoils to varying moisture conditions. This study carried out column leaching experiments on four spoil types, originating from Queensland, Australia. Following characterisation of the fresh spoil material, four moisture regimes were tested: three wetting-drying conditions with leaching occurring biweekly, weekly, and fortnightly and one completely saturated regime with leaching occurring weekly. Thirty-four leaching cycles were conducted except for one spoil type for which only 12 cycles were completed. Results showed higher EC and leachate ion concentrations from the saturated regime, while among the wetting-drying regimes, the spoil leached on a fortnightly basis resulted in higher salt release for two geochemically similar spoil types. Overall, lower and steady pH was recorded for spoils leached under saturated conditions. Irrespective of spoil type, sodium was the dominating cation contributing to the overall leachate salinity. The paper provides new insights into parameterising leaching models and in the role of water-rock interactions which informs experimental design and conceptualisation of full-scale models.

Open cut coal mining produces large amounts of spoil (waste rock) from which salts are leached. Planning for mine closure requires numerical models that can predict salt leaching rates over a range of time scales. This paper describes the modelling of salt leaching from low-sulfide coal mine spoils at two experimental scales: laboratory columns (sample volumes ≈ 0.001 m 3 ) and mesocosms (sample volumes ≈ 1 m 3 ), representing the in-situ spoil properties. Ten spoil samples representing two spoil types and a range of moisture conditions were studied. Simple models with slow and fast reaction rate parameters were derived, and the parameters and their scale factors were estimated empirically using the observations. Using three sample pairs that represent the effect of decreasing particle size, scale factors for the slow reaction rate were estimated to be 7.5, 18.1, and 3.4, and for the fast reaction rate were estimated to be 1.1, 4.4, and 7.4. Using three sample pairs that represent the effect of increasing moisture content, the corresponding scale factors were 1.1, 1.3, 3.0 and 0.92, 0.25, and 3.8. Particle size and flow rate are concluded to be important controls on reaction rate parameters and their upscaling, with moisture content also important but with more complex and variable effects. Experimental limitations prevented other theoretically relevant properties being identified as controls. The results provide new information about potential salt release rates from low-sulfide coal mine spoils. Full-scale applications would require complementary parameterisation of the spoil pile hydrology.

The signature of selection is a crucial concept in evolutionary biology that refers to the pattern of genetic variation which arises in a population due to natural selection. In the context of climate adaptation, the signature of selection can reveal the genetic basis of adaptive traits that enable organisms to survive and thrive in changing environmental conditions. Breeds living in diverse agroecological zones exhibit genetic “footprints” within their genomes that mirror the influence of climate-induced selective pressures, subsequently impacting phenotypic variance. It is assumed that the genomes of animals residing in these regions have been altered through selection for various climatic adaptations. These regions are known as signatures of selection and can be identified using various summary statistics. We examined genotypic data from eight different cattle breeds (Gir, Hariana, Kankrej, Nelore, Ongole, Red Sindhi, Sahiwal, and Tharparkar) that are adapted to diverse regional climates. To identify selection signature regions in this investigation, we used four intra-population statistics: Tajima’s D, CLR, iHS, and ROH. In this study, we utilized Bovine 50 K chip data and four genome scan techniques to assess the genetic regions of positive selection for high-temperature adaptation. We have also performed a genome-wide investigation of genetic diversity, inbreeding, and effective population size in our target dataset. We identified potential regions for selection that are likely to be caused by adverse climatic conditions. We observed many adaptation genes in several potential selection signature areas. These include genes like HSPB2, HSPB3, HSP20, HSP90AB1, HSF4, HSPA1B, CLPB, GAP43, MITF, and MCHR1 which have been reported in the cattle populations that live in varied climatic regions. The findings demonstrated that genes involved in disease resistance and thermotolerance were subjected to intense selection. The findings have implications for marker-assisted breeding, understanding the genetic landscape of climate-induced adaptation, putting breeding and conservation programs into action.

This study seeks a comprehensive exploration of genome-wide selective processes impacting morphometric traits across diverse cattle breeds, utilizing an array of statistical methods. Morphometric traits, encompassing both qualitative and quantitative variables, play a pivotal role in characterizing and selecting livestock breeds based on their external appearance, size, and physical attributes. While qualitative traits, such as color, horn structure, and coat type, contribute to adaptive features and breed identification, quantitative traits like body weight and conformation measurements bear a closer correlation with production characteristics. This study employs advanced genotyping technologies, including the Illumina BovineSNP50 Bead Chip and next-generation sequencing methods like Reduced Representation sequencing, to identify genomic signatures associated with these traits. We applied four intra-population methods to find evidence of selection, such as Tajima’s D, CLR, iHS, and ROH. We found a total of 40 genes under the selection signature, that were associated with morphometric traits in five cattle breeds (Kankrej, Tharparkar, Nelore, Sahiwal, and Gir). Crucial genes such as ADIPDQ, DPP6, INSIG1, SLC35D2 in Kankrej, LPL, ATP6V1B2, CDC14B in Tharparkar, HPSE2, PLAG1 in Nelore, PCSK1, PRKD1 in Sahiwal, and GNAQ, HPCAL1 in Gir were identified in our study. This approach provides valuable insights into the genetic basis of variations in body weight and conformation traits, facilitating informed selection processes and offering a deeper understanding of the evolutionary and domestication processes in diverse cattle breeds.

Electric vertical takeoff and landing (eVTOL) aircraft are employed in several applications such as aerial photography, delivery, search and rescue missions owing to their compactness and ability to hover. However, they inherently have lower endurance and range as compared to their fixed-wing counterparts. Existing approaches to address this issue have explored increasing the mechanical, electrical, or aerodynamic efficiency of the system, replacing the batteries of the drone on ground stations, recharging them in-flight via large wireless chargers or laser power beaming, and several other innovative solutions exist.Given the demand for long-duration flights and long range for applications such as drone delivery and urban aerial mobility (UAM), this dissertation explores various ways to increase the endurance of eVTOLs by adding modularity and incorporating novel design and talks about the interesting challenges that arise from them.We begin by calculating how the flight time of an eVTOL is affected by simply adding more battery. We find that there is a fundamental flight time limit for hovering eVTOL that cannot be exceeded by just adding more battery. This motivates the exploration of new designs and approaches to improve their endurance.First, we explore a simple approach of tethering a series of multirotors to a power source. We evaluate the power requirements to run such a system which helps in design optimization and can be used to guarantee electrical safety. We discover that there exists a critical boundary of thrusts the multirotors can produce that cannot be exceeded due to fundamental electrical limitations. The boundary can be manipulated by changing the voltage of the power supply or the resistance of the cables. We also compare the power consumption for one tethered quadcopter and two tethered quadcopters and show that for large quadcopters far enough from the anchor point, a two-quadcopter system consumes lesser power.Next, we explore the idea of using the multirotor battery in stages to discard the discharged portion of the battery which results in lower power consumption due to reduced mass. We find that even if we stage the energy source continuously (e.g. gasoline in a combustion engine), there still exists a fundamental flight time limit that cannot be exceeded. We consider two optimal staging problems that aim to maximize the flight time and present analytical or visual solutions which are validated experimentally.Then, we present the idea of flying batteries – modular batteries that can fly to a mission quadcopter, dock with it, power and recharge it in-flight, and fly away after discharging, allowing us to repeat the process. This approach lifts the limitation due to energy storage. We present an analysis to evaluate the constraints that need to be satisfied to unlock unlimited endurance for eVTOL aircraft. We then discuss a stochastic model that can be used to predict the probability of success of a long-duration mission when using flying batteries.Lastly, we present a computer vision-based solution that can be used to get the flying batteries system working in the real world. The solution involves using purely onboard sensing for the mission vehicle via a combination of an inertial measurement unit (IMU) and a camera that can generate pose measurements with respect to the flying battery. Measurements from these sensors can be fused to generate relative state estimates that can be used to dock two vehicles in-flight with a centimeter-level precision.We strive to experimentally validate the analysis and proposed approaches presented throughout this thesis.