Increasing Sequence Length Research Articles

Synchronization of Gold sequences based on fast transform in a truncated basis of Walsh–Hadamard functions

Based on the analysis of the structures of isomorphic multiplicative groups of extended Galois fields, it is established that any cyclic shift of a pseudo-random Gold sequence can be transformed into a function belonging to the complete set of analogues of Rademacher functions of the corresponding dimension. This made it possible to develop a new algorithm for fast synchronization of Gold sequences based on the calculation of their discrete convolution using fast spectral transformation in a truncated basis of Walsh–Hadamard functions. The gain of the developed algorithm in terms of the number of arithmetic operations compared to the traditional method of calculating discrete convolution increases with increasing sequence length N and for N=511.1023 is approximately 3.4 times.

NBESM: A Novel BERT Integrated with Sequence Model for Classifying the News in Telugu

News classification helps individuals understand the political and social landscape of their region, enabling them to make informed decisions and participate in civic activities. Accurate classification ensures that readers can distinguish between news, opinion, and other types of content, maintaining the credibility and ethics of journalism. Many people prefer consuming news content in their native language. By classifying news in regional language, media outlets can cater to the language preferences of their audience. Traditional approaches have classified using Bag of Words and embedding approaches but Word embeddings are limited to the vocabulary seen during training. Words that were not present in the training data will have no pre-defined embeddings, and handling such out-of-vocabulary words can be challenging. Later, researchers started experimenting with neural networks such as RNN, GRU and LSTM. While LSTMs are designed to capture long-range dependencies, they can still struggle with very long sequences. As the sequence length increases, LSTMs may face challenges in retaining relevant information over extended periods. The proposed research integrates LSTM with transformers. Transformers excel at capturing contextual relationships across text, while LSTMs are designed to capture sequential patterns. This integration can be particularly useful for tasks that require modeling both local context and long-range dependencies.

Efficient matrix profile computation with Euclidean distance using Eigen transformation: Performance evaluation based on beat-to-beat interval (BBI) data.

The matrix profile serves as a fundamental tool to provide insights into similar patterns within time series. Existing matrix profile algorithms have been primarily developed for the normalized Euclidean distance, which may not be a proper distance measure in many settings. The methodology work of this paper was motivated by statistical analysis of beat-to-beat interval (BBI) data collected from smartwatches to monitor e-cigarette users' heart rate change patterns for which the original Euclidean distance ( -norm) would be a more suitable choice. Yet, incorporating the Euclidean distance into existing matrix profile algorithms turned out to be computationally challenging, especially when the time series is long with extended query sequences. We propose a novel methodology to efficiently compute matrix profile for long time series data based on the Euclidean distance. This methodology involves four key steps including (1) projection of the time series onto eigenspace; (2) enhancing singular value decomposition (SVD) computation; (3) early abandon strategy; and (4) determining lower bounds based on the first left singular vector. Simulation studies based on BBI data from the motivating example have demonstrated remarkable reductions in computational time, ranging from one-fourth to one-twentieth of the time required by the conventional method. Unlike the conventional method of which the performance deteriorates sharply as the time series length or the query sequence length increases, the proposed method consistently performs well across a wide range of the time series length or the query sequence length.

Flood Forecasting Method and Application Based on Informer Model

Flood Forecasting Method and Application Based on Informer Model

A CRISPR-based strategy for targeted sequencing in biodiversity science.

Many applications in molecular ecology require the ability to match specific DNA sequences from single- or mixed-species samples with a diagnostic reference library. Widely used methods for DNA barcoding and metabarcoding employ PCR and amplicon sequencing to identify taxa based on target sequences, but the target-specific enrichment capabilities of CRISPR-Cas systems may offer advantages in some applications. We identified 54,837 CRISPR-Cas guide RNAs that may be useful for enriching chloroplast DNA across phylogenetically diverse plant species. We tested a subset of 17 guide RNAs invitro to enrich plant DNA strands ranging in size from diagnostic DNA barcodes of 1,428 bp to entire chloroplast genomes of 121,284 bp. We used an Oxford Nanopore sequencer to evaluate sequencing success based on both single- and mixed-species samples, which yielded mean chloroplast sequence lengths of 2,530-11,367 bp, depending on the experiment. In comparison to mixed-species experiments, single-species experiments yielded more on-target sequence reads and greater mean pairwise identity between contigs and the plant species' reference genomes. But nevertheless, these mixed-species experiments yielded sufficient data to provide ≥48-fold increase in sequence length and better estimates of relative abundance for a commercially prepared mixture of plant species compared to DNA metabarcoding based on the chloroplast trnL-P6 marker. Prior work developed CRISPR-based enrichment protocols for long-read sequencing and our experiments pioneered its use for plant DNA barcoding and chloroplast assemblies that may have advantages over workflows that require PCR and short-read sequencing. Future work would benefit from continuing to develop invitro and in silico methods for CRISPR-based analyses of mixed-species samples, especially when the appropriate reference genomes for contig assembly cannot be known a priori.

Is Over-parameterization a Problem for Profile Mixture Models?

Biochemical constraints on the admissible amino acids at specific sites in proteins lead to heterogeneity of the amino acid substitution process over sites in alignments. It is well known that phylogenetic models of protein sequence evolution that do not account for site heterogeneity are prone to long-branch attraction (LBA) artifacts. Profile mixture models were developed to model heterogeneity of preferred amino acids at sites via a finite distribution of site classes each with a distinct set of equilibrium amino acid frequencies. However, it is unknown whether the large number of parameters in such models associated with the many amino acid frequency vectors can adversely affect tree topology estimates because of over-parameterization. Here, we demonstrate theoretically that for long sequences, over-parameterization does not create problems for estimation with profile mixture models. Under mild conditions, tree, amino acid frequencies, and other model parameters converge to true values as sequence length increases, even when there are large numbers of components in the frequency profile distributions. Because large sample theory does not necessarily imply good behavior for shorter alignments we explore the performance of these models with short alignments simulated with tree topologies that are prone to LBA artifacts. We find that over-parameterization is not a problem for complex profile mixture models even when there are many amino acid frequency vectors. In fact, simple models with few site classes behave poorly. Interestingly, we also found that misspecification of the amino acid frequency vectors does not lead to increased LBA artifacts as long as the estimated cumulative distribution function of the amino acid frequencies at sites adequately approximates the true one. In contrast, misspecification of the amino acid exchangeability rates can severely negatively affect parameter estimation. Finally, we explore the effects of including in the profile mixture model an additional "F-class" representing the overall frequencies of amino acids in the data set. Surprisingly, the F-class does not help parameter estimation significantly and can decrease the probability of correct tree estimation, depending on the scenario, even though it tends to improve likelihood scores.

Research on multi-model prediction of skeleton curves of prefabricated concrete columns based on Residual fusion Long Short-Term Memory -Transformer

This study aims to investigate a deep learning-based model that permits quick evaluation of the seismic performance of prefabricated concrete columns under bidirectional seismic influences. To capture the time-series properties of the skeleton curves and simultaneously solve the difficulty of predicting the randomness of the initial points of the curves, two models, the fusion model ResMLP-LSTM-Transformer and the single ResMLP, are used to predict the skeleton curves for different concrete columns. For this objective, a skeleton curve dataset of 300 concrete columns with 14 physical properties of each column as input parameters and coordinates of each point of the skeleton curve as output was created. In addition, after utilizing 10-fold cross-validation and hyper-parameter tuning to the models, the predictions of each model were compared with curve results obtained from experiments and finite elements, and the accuracy of the predictions from each model was quantified using RMSE, MSE, MAE, MAPE, and R2. Finally, a thorough investigation was conducted into the impact of the ResMLP sequence length on the fusion model's accuracy. The results demonstrate that despite the diverse and variable physical characteristics of concrete columns, both the fusion model and ResMLP are capable of producing accurate predictions, with the fusion model having a better prediction performance. In addition, the evaluation indexes demonstrate that the fusion model's prediction error increases as ResMLP's sequence length increases, while the RMLT_3 model's prediction performance at a sequence length of three is optimal.

L-SeqSleepNet: Whole-cycle Long Sequence Modelling for Automatic Sleep Staging.

Human sleep is cyclical with a period of approximately 90 minutes, implying long temporal dependency in the sleep data. Yet, exploring this long-term dependency when developing sleep staging models has remained untouched. In this work, we show that while encoding the logic of a whole sleep cycle is crucial to improve sleep staging performance, the sequential modelling approach in existing state-of-the-art deep learning models are inefficient for that purpose. We thus introduce a method for efficient long sequence modelling and propose a new deep learning model, L-SeqSleepNet, which takes into account whole-cycle sleep information for sleep staging. Evaluating L-SeqSleepNet on four distinct databases of various sizes, we demonstrate state-of-the-art performance obtained by the model over three different EEG setups, including scalp EEG in conventional Polysomnography (PSG), in-ear EEG, and around-the-ear EEG (cEEGrid), even with a single EEG channel input. Our analyses also show that L-SeqSleepNet is able to alleviate the predominance of N2 sleep (the major class in terms of classification) to bring down errors in other sleep stages. Moreover the network becomes much more robust, meaning that for all subjects where the baseline method had exceptionally poor performance, their performance are improved significantly. Finally, the computation time only grows at a sub-linear rate when the sequence length increases.

Stability of Characteristics of Physical Random Number Generators

The development of technologies leads to the need of revising the methods used to obtain cryptographic keys. The randomness parameters of sequences generated by physical random sequence generators are affected by the physical parameters of the recording equipment and the environment. The requirements for the randomness of the sequence, when passing the binary test, are obtained. It is shown that as the sequence length increases, the requirements for possible deviations from the equiprobable distribution of 0 and 1 increase. The randomness stability of the sequences generated by the generator based on the study of photons is estimated. The oscillator under study consists of an LED and a compact silicon photomultiplier designed to detect low-power light radiation. Possible physical processes leading to the deterioration of the randomness of sequences are shown. The possibility of using a generator of random numerical sequences based on a small-sized silicon photomultiplier for cryptographic purposes is estimated.

Joint Compensation of TX/RX IQ Imbalance and Channel Parameters for OTSM Systems

Orthogonal time sequency multiplexing (OTSM), robust to doubly-selective fading channel, performs similar to orthogonal time frequency space (OTFS) system with less modulation/demodulation complexity. Direct conversion OTSM system suffers from hardware impairments such as in-phase and quadrature phase (IQ) imbalance at the transmitter (TX) and receiver (RX). At high frequencies, hardware imperfections are inevitable and should be compensated for better performance. In this Letter, we propose joint estimation of TX and RX IQ imbalance and estimate channel parameters in the delay time domain using pseudo noise based training sequence approach. The mean square error of the estimated channel parameters reduce with increase in training sequence length and power. With matched filtered Gauss-Seidel iterative method of detection of low complexity, it is shown that the compensated system performs close to ideal system under fractional delay and fractional Doppler.

Prediction of PGA in earthquake early warning using a long short-term memory neural network

SUMMARYPeak ground acceleration (PGA) is a key parameter used in earthquake early warning systems to measure the ground motion strength and initiate emergency protocols at major projects. The traditional P-wave peak displacement-dependent PGA prediction model (Pd-PGA model) tends to underestimate the PGA for large earthquakes because it cannot make full use of the fault continuity rupture information hidden in the time-varying process of ground motion. In this paper, a continuous PGA prediction long short-term memory (LSTM) neural network model is proposed. The model takes eight sequential features of stations that are proxies of the energy and other physical parameters as input and provides the recorded PGA at the station as the target output. A total of 5961 records from 119 earthquakes recorded by the Japanese Strong-Motion Earthquake Network (K-NET) in Japan are used to train the neural network and 3433 records from 73 earthquakes are used as the test set to verify the model’s generalization ability. The results show that within the same data set, the residuals of the predicted PGA for the proposed model are smaller than those of the Pd-PGA model and that the problem of PGA underestimation is resolved. The prediction accuracy also improves with increasing sequence length, which indicates that the LSTM neural network learns the rules hidden in the time series. To further verify the model’s generalization ability, the model performance is analyzed for an M 7.3 earthquake that was not included in the training or test data sets. The results show that the residuals of the predicted PGA for the event are consistent with those for the test data set, indicating that the model has good generalization ability.

An Effective Self-Attention-Based Hybrid Model for Short-Term Traffic Flow Prediction

Vehicle exhaust is one of the main sources of carbon emissions. The short-term traffic flow prediction plays an important role in alleviating traffic congestion, optimizing the travel structure, and reducing traffic carbon emissions. The current advanced models of short-term traffic flow prediction are evaluated in this work, especially their inadequacies. To improve the prediction accuracy and ensure fine traffic management, an effective self-attention-based hybrid model is proposed to predict the short-term traffic flow. The proposed model includes an encoder-decoder neural network module and a self-attention mechanism module. The self-attention mechanism module is applied as a feature extraction unit in this hybrid model to enhance the ability of key information capture and to settle the problem on key information disappearing due to the increasing sequence length in traditional models. The dataset of the Guangdong freeway toll station is used for the experimental testing. Compared with several baseline models, the proposed model is more suitable for real-time prediction and can provide highly accurate results. Also, a better interpretability is presented in this proposed model. The experimental results showed that MAE, RMSE, and MAPE of the proposed model are 3.01, 4.38, and 12.99%, respectively. Our new hybrid model gives a higher accuracy than the support vector regression (SVR) model, LSTM neural network-attention (LSTM-attention) model, and temporal convolutional network (TCN) model. It shows that the proposed model in this work is favorable to the short-term traffic flow prediction.

О точности нормальной аппроксимации для распределения числа повторений в стационарной дискретной случайной последовательности

The paper presents the problem of asymptotic normality of the number of r-fold repetitions of characters in a segment of a (strictly) stationary discrete random sequence on the set {1, 2,..., N} with the uniformly strong mixing property. It is shown that in the case when the uniformly strong mixing coefficient φ(t) for an arbitrarily given α > 0 decreases as t-6-α, then the distance in the uniform metric between the distribution function of the number of repetitions and the distribution function of the standard normal law decreases at a rate of O(n- δ) with increasing sequence length n for any α ∈ (0; α (32 + 4α )-1)).

Machine learning overcomes human bias in the discovery of self-assembling peptides.

Peptide materials have a wide array of functions, from tissue engineering and surface coatings to catalysis and sensing. Tuning the sequence of amino acids that comprise the peptide modulates peptide functionality, but a small increase in sequence length leads to a dramatic increase in the number of peptide candidates. Traditionally, peptide design is guided by human expertise and intuition and typically yields fewer than ten peptides per study, but these approaches are not easily scalable and are susceptible to human bias. Here we introduce a machine learning workflow-AI-expert-that combines Monte Carlo tree search and random forest with molecular dynamics simulations to develop a fully autonomous computational search engine to discover peptide sequences with high potential for self-assembly. We demonstrate the efficacy of the AI-expert to efficiently search large spaces of tripeptides and pentapeptides. The predictability of AI-expert performs on par or better than our human experts and suggests several non-intuitive sequences with high self-assembly propensity, outlining its potential to overcome human bias and accelerate peptide discovery.

Sequence Length of HIV-1 Subtype B Increases over Time: Analysis of a Cohort of Patients with Hemophilia over 30 Years.

We aimed to investigate whether the sequence length of HIV-1 increases over time. We performed a longitudinal analysis of full-length coding region sequences (FLs) during an HIV-1 outbreak among patients with hemophilia and local controls infected with the Korean subclade B of HIV-1 (KSB). Genes were amplified by overlapping RT-PCR or nested PCR and subjected to direct sequencing. Overall, 141 FLs were sequentially determined over 30 years in 62 KSB-infected patients. Phylogenetic analysis indicated that within KSB, two FLs from plasma donors O and P comprised two clusters, together with 8 and 12 patients with hemophilia, respectively. Signature pattern analysis of the KSB of HIV-1 revealed 91 signature nucleotide residues (1.1%). In total, 48 and 43 signature nucleotides originated from clusters O and P, respectively. Six positions contained 100% specific nucleotide(s) in clusters O and P. In-depth FL analysis for over 30 years indicated that the KSB FL significantly increased over time before combination antiretroviral therapy (cART) and decreased with cART. This increase occurred due to the significant increase in env and nef genes, originating in the variable regions of both genes. The increase in sequence length of HIV-1 over time suggests an evolutionary direction.

An efficient methodology for aspect-based sentiment analysis using BERT through refined aspect extraction

Aspect-Based Sentiment Analysis (ABSA) has become a trending research domain due to its ability to transform lives as well as the technical challenges involved in it. In this paper, a unique set of rules has been formulated to extract aspect-opinion phrases. It helps to reduce the average sentence length by 84% and the complexity of the text by 50%. A modified rank-based version of Term-Frequency - Inverse-Document-Frequency (TF-IDF) has been proposed to identify significant aspects. An innovative word representation technique has been applied for aspect categorization which identifies both local as well as global context of a word. For sentiment classification, pre-trained Bidirectional Encoder Representations from Transformers (BERT) has been applied as it helps to capture long-term dependencies and reduce the overhead of training the model from scratch. However, BERT has drawbacks like quadratic drop in efficiency with an increase in sequence length which is limited to 512 tokens. The proposed methodology mitigates these drawbacks of a typical BERT classifier accompanied by a rise in efficiency along with an improvement of 8% in its accuracy. Furthermore, it yields enhanced performance and efficiency compared to other state-of-the-art methods. The assertions have been established through extensive analysis upon movie reviews and Sentihood data-sets.

Identification of Single Nucleotide Polymorphisms and Analysis of Linkage Disequilibrium in Different Bamboo Species Using the Candidate Gene Approach

Bamboos are one of the most beautiful and useful plants on Earth. The genetic background and population structure of bamboos are well known, which helps accelerate the process of artificial domestication of bamboo. Partial sequences of six genes involved in nitrogen use efficiency in 32 different bamboo species were analyzed for occurrence of single nucleotide polymorphisms (SNPs). The nucleotide diversity θw and total nucleotide polymorphisms πT of the sequenced DNA regions was 0.05137 and 0.03332, respectively. Both πnonsyn /πsyn and Ka/Ks values were 0.1. Estimation and analysis of linkage disequilibrium of five genes revealed that with the increase in nucleotide sequence length, the degree of SNP linkage disequilibrium decreased rapidly. We inferred the population genetic structure of 32 bamboo species based on the SNP loci of six genes with frequencies >18%. 32 bamboo species were divided into five categories, which indicated that the combined population of all bamboo species had obvious multivariate characteristics and was heterogeneous; red (Group 1) and green (Group 2) were the main groups.

Optimizing imitation: Examining cognitive factors leading to imitation, overimitation, and goal emulation in preschoolers

Humans imitate patently irrelevant actions known as overimitation, and rather than decreasing with age, overimitation increases with age. Whereas most overimitation research has focused on social factors associated with overimitation, comparatively little is known about the cognitive- and task-specific features that influence overimitation. Specifically, developmental contrasts between imitation and overimitation are confounded by the addition of irrelevant actions to causally necessary actions, increasing sequence length, cognitive load, and processing costs—variables known to be age dependent. We constructed a novel puzzle box task such that a four-step imitation, four-step overimitation, and two-step efficient sequence could be demonstrated using the same apparatus on video. In Experiments 1 and 2, 2.5- to 5-year-olds randomly assigned to imitation and overimitation groups performed significantly more target actions than baseline control groups. Rates of imitation and overimitation increased as a function of age, with older preschoolers outperforming younger preschoolers in both conditions. In Experiment 3, preschoolers were shown a video of an efficient two-step demonstration prior to testing. After they responded, they were shown a four-step overimitation video and were tested on the same puzzle box. Children imitated the efficient demonstration, but after watching the overimitation video, they also overimitated the irrelevant actions. Once again, older children overimitated more than younger children. Together, results show that preschoolers are faithful, flexible, and persistent overimitators. The fidelity and flexibility of overimitation are constrained not only by social factors but also by basic cognitive processes that vary across age groups. As these constraints diminish, overimitation and flexible (optimal) imitation increases.

Design and Optimization of Tropomyosin Fragments for Tropomodulin Interaction Studies

Monomeric actin, or G-actin, polymerizes into filamentous actin, or F-actin which is an important component of cellular structures controlling many processes including vesicle trafficking, cell motility, inter and intracellular transport, and muscle contraction. Actin filaments are composed of a pointed end (slow growing) and a barbed end (fast growing). Interactions at each end enables control of the length of actin filaments. Tropomodulin (Tmod1-4) modulates actin filament length by interacting with tropomyosin (Tpm) and actin at the pointed end of the filament. Tmod binds two Tpm molecules through two distinct Tpm binding sites. Establishing the structure of each Tmod-Tpm binding interface is critical to understand the thin filament length regulation mechanism of Tmod. To facilitate NMR structure determination, we have established the optimal size of Tpm fragment necessary to form respective stable complexes. The four fragments we designed were residues 1-43, 1-33, 1-26 or 1-19 of Tpm1.8 fused with a GCN4 sequence promoting coiled-coil formation. We used NMR to determine and compare spectral effects between the optimized Tpm fragments and Tpm-binding sites of Tmod2. A significant change in spectra occurred when length of Tpm sequence changed from 19 to 26 residues and with almost no change with further increase in Tpm sequence length. Stability of the designed Tpm fragments and their affinity for Tpm-binding sites on Tmod2 were evaluated by CD spectroscopy. Our results revealed that Tpm1.8 fragments containing residues 1-19 and 1-43 were more stable with a melting temperature of 44 and 52oC, respectively, however, the 1-19 fragment was not long enough to represent the whole binding interface. The calculated dissociation constants showed strong affinity of Tmod2 for the Tpm1.8 1-43. Therefore, Tpm1.8-1-43 fragment is the most optimal for use in NMR studies to further study its interaction with Tmod.

Towards Non-Saturating Recurrent Units for Modelling Long-Term Dependencies

Modelling long-term dependencies is a challenge for recurrent neural networks. This is primarily due to the fact that gradients vanish during training, as the sequence length increases. Gradients can be attenuated by transition operators and are attenuated or dropped by activation functions. Canonical architectures like LSTM alleviate this issue by skipping information through a memory mechanism. We propose a new recurrent architecture (Non-saturating Recurrent Unit; NRU) that relies on a memory mechanism but forgoes both saturating activation functions and saturating gates, in order to further alleviate vanishing gradients. In a series of synthetic and real world tasks, we demonstrate that the proposed model is the only model that performs among the top 2 models across all tasks with and without long-term dependencies, when compared against a range of other architectures.