Sequential Machine Research Articles

Data-driven surrogate model for wind turbine damage equivalent load

Abstract. Aeroelastic simulations are employed to assess wind turbines in accordance with IEC standards in the time domain. These analyses enable the evaluation of fatigue and extreme loads experienced by wind turbine components. Such simulations are essential for several reasons, including but not limited to reducing safety margins in wind turbine component design by accounting for a wide range of uncertainties in wind and wave conditions and fulfilling the requirements of the digital twin, which necessitates a comprehensive set of simulations for calibration. Thus, it is essential to develop computationally efficient yet accurate models that can replace costly aeroelastic simulations and data processing. To address this challenge, we propose a data-driven approach to construct surrogate models for the damage equivalent load (DEL) based on aeroelastic simulation outputs. Our method provides a quick and efficient way to calculate DEL using wind input signals without the need for time-consuming aeroelastic simulations. Our study focuses on utilizing a sequential machine learning (ML) method to map wind speed time series to DEL. Additionally, we demonstrate the versatility of the developed and trained surrogate models by testing them on a wind turbine in the wake and applying transfer learning to enhance their predictive accuracy.

A Deep Learning Approach for Predicting Aerial Suppressant Drops in Wildland Firefighting Using Automatic Dependent Surveillance–Broadcast Data

This study utilizes Automatic Dependent Surveillance–Broadcast (ADS-B) data sourced by the OpenSky Network to curate a dataset aimed at enhancing the precision of aerial suppressant drop predictions in wildland firefighting. By amalgamating ADS-B data with Automated Telemetry Unit (ATU) drop information, this research constructs a reliable base for analyzing the spatial aspects of aerial firefighting operations. Using sequential machine learning models, specifically Long Short-Term Memory (LSTM) networks and 1D Convolutional Neural Networks (1DCNN), the study interprets complex flight dynamics to predict drop locations. The dataset, covering 2017 to 2023, is labeled and segmented to reflect accurate suppressant release events, facilitating the distinction between drop and non-drop activities in fixed-wing aircraft. The LSTM model demonstrated strong predictive performance with an F1 score of 0.922, effectively identifying suppressant drop events with high accuracy. This model’s reliable predictions can significantly improve situational awareness in real-time aerial firefighting operations, enabling more informed decision-making and better coordination of resources during wildfire events.

On decision problems concerning contextual insertions and deletions

The notions of stability, anti-stability, and error-correctability of a language that is modified by making contextual insertions in the words of the language were introduced in a previous paper by Bottoni et al. in 2011, where it was shown that these properties are decidable for regular languages. The authors proposed investigating the decidability of these properties for other classes of languages. Here, we derive necessary and sufficient conditions for a class of languages to have decidable stable, anti-stable, and error-correctable properties, and use these conditions to exhibit general classes of languages (strictly greater than the regular languages) for which the properties are decidable, and also simple classes (the first such classes) for which the properties are undecidable. We obtain identical results for the case when contextual deletions (instead of insertions) are made in the words of the language, and also with mixes of insertions and deletions. Our constructions also demonstrate that certain general problems involving nondeterministic generalized sequential machines (GSMs) applied to languages accepted by deterministic machine models are decidable, which is surprising as the deterministic language families do not need to be closed under GSM mappings.

Exploiting Temporal Features in Calculating Automated Morphological Properties of Spiky Nanoparticles Using Deep Learning

Object segmentation in images is typically spatial and focuses on the spatial coherence of pixels. Nanoparticles in electron microscopy images are also segmented frame by frame, with subsequent morphological analysis. However, morphological analysis is inherently sequential, and a temporal regularity is evident in the process. In this study, we extend the spatially focused morphological analysis by incorporating a fusion of hard and soft inductive bias from sequential machine learning techniques to account for temporal relationships. Previously, spiky Au nanoparticles (Au-SNPs) in electron microscopy images were analyzed, and their morphological properties were automatically generated using a hourglass convolutional neural network architecture. In this study, recurrent layers are integrated to capture the natural, sequential growth of the particles. The network is trained with a spike-focused loss function. Continuous segmentation of the images explores the regressive relationships among natural growth features, generating morphological statistics of the nanoparticles. This study comprehensively evaluates the proposed approach by comparing the results of segmentation and morphological properties analysis, demonstrating its superiority over earlier methods.

Binary distinguishability operation

This paper analyzes a new distinguishability operation for finite deterministic machines and languages. The research was inspired by the “Gedanken experiments” on sequential machines performed by Moore in 1956, and extends the study of the unary distinguishability operation to a binary one. Besides studying the new operation's properties, we give a tight bound of its state complexity on regular languages, including the case for finite languages.

Design and Implementation of a Reconfigurable Corrective Control System Subject to Permanent Faults in the Controller.

This article presents a reconfiguration strategy for the corrective controller achieving model matching control of an input/state asynchronous sequential machine (ASM). The considered controller is vulnerable to permanent faults that degenerate a subset of the controller's states. If the controller has a certain amount of redundancy in terms of its states, one can build a reconfiguration scheme in which the functionality of degenerated states is taken over by supplementary states. The proposed reconfiguration scheme is superior to conventional methods of fault tolerance with hardware redundancy since the required number of redundant states is much smaller. Hardware experiments on field-programmable gate array (FPGA) circuits are provided to validate the applicability of the proposed scheme. The present study serves as the first research report on the reconfigurable corrective controller.

Mutual modeling of sequential and parallel word computations

The work is devoted to the connection between parallel and sequential computing. On the one hand, we consider a class of word predicates based on sequential calculations, limited in memory by constants and having polynomial time complexity. On the other hand, we consider a class of word predicates that are computable on parallel alternating machines in logarithmic time. The coincidence of the corresponding classes is proven. The direction of using the obtained results for mutual transformation and combination of calculations on molecular biosimilar sequential machines and parallel calculations on vector-matrix computers is proposed. Intended applications: real-time image processing for control tasks, analysis of large texts and other big data.

Input/output corrective control of switched asynchronous sequential machines under arbitrary switching

This article presents a novel modeling and corrective control of switched asynchronous sequential machines (ASMs) with input/output submachines. In particular, the considered switched ASM is harnessed by the external switching signal that provokes arbitrary change of the mode or submachine. The existence of the external switching signal imposes unpredictable state drift opposite to model matching as well as ambiguity in determining the active submachine. We first present a state observer for state observation and identification of the active submachine. Based on the information delivered by the observer, we address the existence condition and design procedure for a corrective controller that matches the stable-state behavior of the closed-loop system to that of a reference model for every possible switching sequence. To demonstrate the validity and applicability of the proposed control scheme, we conduct hardware experiments on field-programmable gate array (FPGA) circuits for a space-borne digital system and provide convincing experimental results.

Leveraging machine learning to dissect role of combinations of amino acids in modulating the effect of zinc on mammalian cell growth.

Although the contributions of individual components of cell culture media are largely known, their combinatorial effects are far less understood. Experiments varying one component at a time cannot identify combinatorial effects, and analysis of the large number of experiments required to decipher such effects is challenging. Machine learning algorithms can help in the analysis of such datasets to identify multi-component interactions. Zinc toxicity in vitro is known to change depending on amino acid concentration in the extracellular medium. Multiple amino acids are known to be involved in this protection. Thirty-two amino acid compositions were formulated to evaluate their effect on the growth of CHO cells under high zinc conditions. A sequential machine learning analysis methodology was used, which led to the identification of a set of amino acids (threonine, proline, glutamate, aspartate, asparagine, and tryptophan) contributing to protection from zinc. Our results suggest that a decrease in availability of these set of amino acids due to consumption may affect cell growth in media formulated with high zinc concentrations, and in contrast, normal levels of these amino acids are associated with better tolerance to high zinc concentration. Our sequential analysis method may be similarly employed for high throughput medium design and optimization experiments to identify interactions among a large number of cell culture medium components.

Design of Smart Traffic Signal using Verilog

Traffic light control can be designed as synchronous sequential machine with finite number of states. Explicit finite state model is used to design the necessary coding for control system using Verilog HDL. The machine is modeled with only six states necessary delay is provided and for that particular delay the necessary traffic lights are set ON and OFF. For illustration just only two roads Chosen and control algorithm controls the traffic lights of that roads proposes a flexible framework which provides a particular delaying particular using click divider, also discusses the issues of modeling the state machine in a synthes is friendly manner. The proposed system aims to optimize traffic flow and reduce congestion by dynamically adjusting signal timings based on real-time traffic data. The design includes modules for vehicle detection, data processing, and signal control. By utilizing Verilog's capabilities, the system achieves efficient and accurate signal management, enhancing over all traffic management and contributing to safer and more stream lined urban mobility..

Hemogram Derived Clinical Traits Evinced Differential Contribution to Explained Variance of Mortality Risk Model at Index Hospitalization for Sars-Cov-2 Infection across Wuhan/D614G, Delta, Omicron and Omicron BA.x Variants

Introduction SARS-CoV-2 infection promulgates an immune dysregulation that drives an inflammatory cascade therefore inducingmorbidity and mortality across the COVID-19 presentation spectrum. Hemogram (CBC) derived traits associated with COVID-19 severity have been reported (Meizlish ML et al., 2021, Dhinata KS et al., 2021), but their relative impact across SARS-CoV-2 variants is unclear. We employed sequential machine learning approachesto identify significantly contributing traits with attendant portion of explained variance (EV%) in hospital mortality risk models across predominant SARS-CoV-2 variants. Methods Our retrospective design, analyses and interpretations followed constructs detailed in the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline. Adult patients with laboratory confirmed COVID-19 who underwent index hospitalization during Wuhan/D614G (March 14, 2020- June 18, 2021), Delta (June 19, 2021- December 18, 2021), Omicron (December 19-March 30, 2022) or Omicron BA.x (March 31, 2022 - April 14, 2023) variant waves at an 820-bed academic public health trust hospital in Florida were studied. Demographics, laboratory results (within 24h of presentation), ICD-10-CM-based comorbidity, COVID-19 directed treatment and administrative data were extracted from electronic medical records under IRB exemption. Generalized regression with adaptive LASSO identified hemogram-derived traits significantly associated with mortality in at least one of the variants while controlling for presentation vital signs, age, sex, extant comorbidities and ultimate COVID-19 directed treatment. Boosted Tree modeling computed within-variant proportion contributed by each trait to model's accuracy (R 2) representing EV% of mortality risk. Traits contributing to one or more variants for at least 1% portion of EV were retained in model. A Bonferroni corrected two-tailed p < .0125 was considered significant. Results We included 6490 consecutively discharge patients distributed across Wuhan/D614G (n=2249), Delta (n=1196), Omicron (n=953) and Omicron BA.x (n=2092) variants with respective mortality of 9.3%, 13.0%, 4.4%, and 2.4% (p< .0001). Patient characteristics and on-presentation laboratory medicine clinical traits are in Table 1. Table 2 provides ranked contribution for each of the 7 traits retained in the model with attendant presentation mortality risk accuracy (R 2) across variant waves. Potent synergy can be generated with biomarkers ratios, as seen with Segmented neutrophil/monocyte ratio (SMR) and lymphocyte/monocyte ratio (LMR), compared to studying these variables individually. Conclusion Sequential machine-learning approaches identified differential expression intensity across SARS-CoV-2 variants in presentation hemogram-derived traits associated with hospital mortality risk modeling at index infection. These findings suggest that COVID-19 presentation risk and level of care assessment tools employing hemogram-derived traits may not generalize across pandemic variants. Platelets counts were associated with mortality in Delta, Wuhan/D614G and Omicron variants. Although it was also associated in Omicron Bx.5, the R 2 is lower, this can be explained because in this study the variant of Omicron Bx.5 had the lowest mortality rate. Studies have shown that platelet activation generates a cascade with more thromboxane A2 and platelet thrombofactor A which increases prothrombotic states (Dhinata KS et al., 2021). The absence of eosinophils was seen in Wuhan/D614G and Delta, the cause of which cannot be ascertained. A theory is that the inflammation caused by SARS-CoV-2 infection releases certain cytokines that induce eosinophil apoptosis (Dhinata KS et al., 2021). Immature neutrophils evidenced across variants can be explained by the inflammatory cascade inducing G-CSF and IL-8 production, which induces neutrophil activation (Meizlish ML et al., 2021).Lymphopenia seen throughout variants contributes to an increase in LMR. One presumption is that the SARS-CoV-2 affects T-Lymphocytes through ACE-2 receptor and CD117 spike proteins (Meizlish ML et al., 2021). Overall, our study brings to attention how hemogram-derived traits express differently across SARS-CoV-2 variants, which could guide the stratification of COVID-19 severity and assist in the care of patients before decompensation is noticed.

Life satisfaction for adolescents with developmental and behavioral disabilities during the COVID-19 pandemic.

This study aimed to identify contextual factors associated with life satisfaction during the COVID-19 pandemic for adolescents with mental, emotional, behavioral, and developmental (MEBD) disabilities. Data were collected from a sample of 1084 adolescents aged 11-21 years from April 2020 to August 2021. This cross-sectional study used a sequential machine learning workflow, consisting of random forest regression and evolutionary tree regression, to identify subgroups of adolescents in the Environmental influences on Child Health Outcomes (ECHO) consortium who demonstrated enhanced vulnerability to lower life satisfaction as described by intersecting risk factors, protective factors, and MEBD disabilities. Adolescents with a history of depression, anxiety, autism, and attention-deficit/hyperactivity disorder were particularly susceptible to decreased life satisfaction in response to unique combinations of stressors experienced during the COVID-19 pandemic. These stressors included decreased social connectedness, decreased family engagement, stress related to medical care access, pandemic-related traumatic stress, and single-caregiver households. Findings from this study highlight the importance of interventions aimed specifically at increasing adolescent social connectedness, family engagement, and access to medical support for adolescents with MEBD disabilities, particularly in the face of stressors, such as a global pandemic. Through a machine learning process, we identified contextualized risks associated with life satisfaction among adolescents with neurodevelopmental disabilities during the COVID-19 pandemic. The COVID-19 pandemic resulted in large-scale social disruptions for children and families. Such disruptions were associated with worse mental health outcomes in the general pediatric population, but few studies have examined specific subgroups who may be at heightened risk. We endeavored to close that gap in knowledge. This study highlights the importance of social connectedness, family engagement, and access to medical support as contributing factors to life satisfaction during the COVID-19 pandemic for adolescents with neurodevelopmental disabilities.

Resilient Corrective Control of Asynchronous Sequential Machines Against Intermittent Loss of Actuator Outputs.

This article proposes a resilient corrective control scheme for input/state asynchronous sequential machines (ASMs) against a class of actuator faults in which certain actuator outputs cannot be generated temporarily. We first present a mathematical formulation to describe the reachability of the controlled ASM damaged by the intermittent loss of actuator outputs. Based on the mathematical formulation, we address the existence condition and design procedure for a state-feedback corrective controller and a diagnoser that achieve resilience, that is, to make the closed-loop system exhibit normal input/state behaviors despite the intermittent loss of actuator outputs. To validate the applicability of the proposed concept and methodology, the closed-loop system of a practical asynchronous digital system is implemented on a field-programmable gate array (FPGA) and experimental verifications are provided.

Sequential Machine Learning for Activity Sequence Prediction from Daily Work Report Data

Sequential Machine Learning for Activity Sequence Prediction from Daily Work Report Data

Teaching to Learn: Sequential Teaching of Learners with Internal States

In sequential machine teaching, a teacher’s objective is to provide the optimal sequence of inputs to sequential learners in order to guide them towards the best model. However, this teaching objective considers a restricted class of learners with fixed inductive biases. In this paper, we extend the machine teaching framework to learners that can improve their inductive biases, represented as latent internal states, in order to generalize to new datasets. We introduce a novel framework in which learners’ inductive biases may change with the teaching interaction, which affects the learning performance in future tasks. In order to teach such learners, we propose a multi-objective control approach that takes the future performance of the learner after teaching into account. This framework provides tools for modelling learners with internal states, humans and meta-learning algorithms alike. Furthermore, we distinguish manipulative teaching, which can be done by effectively hiding data and also used for indoctrination, from teaching to learn which aims to help the learner become better at learning from new datasets in the absence of a teacher. Our empirical results demonstrate that our framework is able to reduce the number of required tasks for online meta-learning, and increases independent learning performance of simulated human users in future tasks.

A Stock Price Prediction Method based on LSTM and K-Means

Artificial neural network, as a nonlinear mapping or adaptive power system made by linking neurons,which can effectively resolve problems such as gradient explosion in stock price prediction processing. Recurrent neural network (RNN) is a common model for processing stock time-series data and is suitable for stock data involving sequential machine learning tasks, but the prediction results are poor when using long time span or nonlinear data for prediction. To address the problems of low prediction accuracy of ordinary neural networks for stock data with poor linearity and the inability of a single LSTM model to show the recommendation level of a target stock, the paper proposes a deep learning factor integrated prediction model based on LSTM-K-Means.On this basis, a stock price prediction method based on long-term and short-term memory network namely LSTM and K-means clustering algorithm is proposed. The method is not only designed to model stock ups and downs at different levels of combinations, but also more intuitively identifies stocks with better ups through returns and volatilities. Through experimental verification, the stock price prediction method based on LSTM with K-Means proposed in this paper is effective.

Self-Repairing Corrective Control for Input/Output Asynchronous Sequential Machines With Transient Faults

Fault-tolerant corrective control for input/output asynchronous sequential machines (ASMs) is studied in this article. In the present problem setting, not only the controlled ASM but also the corrective controller itself is vulnerable to transient faults. The fault outcome is aggravated when the transient fault occurs to the controller, since the adverse effect of the fault is propagated toward the controlled ASM. We first address a scheme of diagnosing transient faults by designing the state observer that induces the resultant faulty state based on the output burst. We then present the existence condition and design algorithm for a corrective controller that recovers the normal behavior of the closed-loop system against transient faults at the ASM, while achieving self-repair against those faults occurring to the controller. Hardware experiments on field-programmable gate array (FPGA) circuits are provided to validate the applicability of the proposed control methodology.

Model matching of input/state switched asynchronous sequential machines with the external switching signal

This article investigates the input/state model matching problem of switched asynchronous sequential machines (ASMs) with the external switching signal in the framework of corrective control theory. The considered switched ASM is governed by the external switching signal that arbitrarily changes the mode of the switched ASM, prompting the active submachine into unpredictable state drift. We address the existence condition and design procedure for a proper corrective controller that matches the stable-state behavior of the closed-loop system to that of a reference model under any switching sequence. An illustrative example is provided for demonstrating the synthesis procedure of the proposed corrective controller. Compared with the prior work, the subject of this study is more challenging since rather than manipulated in favor of attaining model matching, the switching signal inflicts adversarial dynamic constraint that must be overcome by the controller.

Boosted Gaze Gesture Recognition Using Underlying Head Orientation Sequence

People find it challenging to control smart systems with complex gaze gestures due to the vulnerability of eye saccades. Instead, the existing works achieved good recognition accuracy of simple gaze gestures because of sufficient eye gaze points but simple gaze gestures have limited applications compared to complex gaze gestures. Complex gaze gestures need a composition of multiple subunits of eye fixation to contain a sequence of gaze points that are clustered and rotated with an underlying complex head orientation relationship. This paper proposes a new set of eye gaze points and head orientation angles as new sequences to recognize complex gaze gestures. Eye gaze points and head orientation angles have a powerful influence on gaze gesture formation. The new sequence was obtained by aligning clustered gaze points and head orientation angles with a simple moving average (SMA) to denoise and interpolate the gap between successive eye fixations. The aligned new sequence of complex gaze gestures was utilized to train sequential machine learning (ML) algorithms. To evaluate the performance of the proposed method, we recruited and recorded the eye gaze and head orientation features of ten participants using an eye tracker. The results show that Boosted Hidden Markov Models (HMM) using Random Subspace methods achieved the best accuracies of 94.72% and 98.1% for complex, and simple gestures respectively, which outperformed the conventional methods.

Reinforcement learning for automatic detection of effective strategies for self-regulated learning

Self-regulated learning (SRL) is an essential skill for achieving one's learning goals, particularly in Digital Learning Environments (DLEs) where system support is often limited compared to traditional classroom settings. However, research has found that learners often struggle to adapt their behaviour to the self-regulatory demands of DLEs. Furthermore, existing SRL analysis tools have limited utility for real-time or individualized prescriptive support of a learner's SRL strategy during a study session.In response to these challenges, we propose a novel approach using reinforcement learning as a framework to optimize the sequence of SRL processes for a learning task. This framework allows us to model and optimize the SRL strategy as a sequential decision-making problem, where each decision corresponds to an SRL process. The goal is to find an optimal sequence of decisions that maximizes performance in learning outcomes such as assessment score or learning gains.We compare the performance of our reinforcement learning framework with other sequential machine learning tools, such as Long Short-Term Memory (LSTM) neural networks and Genetic Algorithms (GA). The results of our study show that our reinforcement learning model outperforms GA and LSTM models in optimizing SRL strategy.The contributions of this work can facilitate the development of a tool which can detect sub-optimal SRL strategy in real-time and enable individualized SRL focused scaffolding.