Identification Task Research Articles

Greek-Cypriot learners’ perception and production of L2 English vowels

Acquiring L2 segments is particularly challenging for L2 learners, especially when the L1 and L2 inventories involve different contrasts and acoustic cues. The present research investigated the perception and production of L2 English vowels by adult Greek-Cypriot learners and examined the effects of orthographic cues and consonantal context in their performance. Perceptual performance was assessed through a forced-choice identification task and production performance through a wordlist-reading and an elicitation task, both analyzed acoustically and through intelligibility ratings. The findings showed the influence of the L1 on both the perception and production of L2 segments, supporting the assumptions of current models of speech perception and production. Learners faced challenges in perceiving the members of L2 contrasts and mostly used their L1 articulatory routines in their productions of L2 vowels. Orthographic cues or consonantal context did not significantly affect learners’ productions or overall perception, although strong contextual effects were observed in individual target vowels. This study is the first to provide an in-depth comparison of CYG and English vowels as well as an examination of the acquisition of L2 English vowels by this population, which can guide EFL teachers and course developers in developing appropriate EFL curricula that incorporate pronunciation instruction.

Transfer-AE: A novel autoencoder-based impact detection model for structural digital twin

Accurately detecting the location and intensity of impacts is crucial for ensuring structural safety. Currently, AI-based structural impact detection methods are widely used for their excellent detection accuracy. However, their generalization capability is limited by the scenarios present in the training data. Many complex and dangerous impact scenarios are difficult to conduct real-world experiments on to collect sufficient samples. To capture all impact scenarios and fully leverage the advantages of AI-based detection technologies, advanced methods involve combining real-world structural monitoring data with corresponding numerical models to construct digital twins. These methods continuously refine the created numerical models with limited real-world data and provide diverse impact scenarios through numerical model simulations. However, there are inevitable differences between digital models and physical models that are challenging to correct through mechanical means. This discrepancy in data distribution between the two models significantly hinders the application of digital twin technology in impact/event identification tasks. To address this challenge, this study proposes a novel model based on autoencoders, named Transfer-AE. Transfer-AE encodes the common features of digital twins in the latent space to bridge the uncertainty gap at a macro scale between numerical models and physical models and synchronously fits the magnitude and location of the impact load in the decoder. This enables consistent detection results for the same impact event, whether the sample comes from the numerical model or the physical model. Transfer-AE includes two operating modes: Mode 1 has a fixed computational complexity with stable inference speed, but the training cost and difficulty increase with data distribution. Mode 2's computational complexity increases with data distribution, but it has a fixed training cost and speed. In both cases involving the geodesic dome structure simulating a deep space habitat and the IASC-ASCE benchmark structure, Transfer-AE demonstrated the best performance in impact localization and quantification tasks compared to mainstream domain-adaptive transfer models.

Digital Dental Biometrics for Human Identification Based on Automated 3D Point Cloud Feature Extraction and Registration.

Intraoral scans (IOS) provide precise 3D data of dental crowns and gingival structures. This paper explores an application of IOS in human identification. We propose a dental biometrics framework for human identification using 3D dental point clouds based on machine learning-related algorithms, encompassing three stages: data preprocessing, feature extraction, and registration-based identification. In the data preprocessing stage, we use the curvature principle to extract distinguishable tooth crown contours from the original point clouds as the holistic feature identification samples. Based on these samples, we construct four types of local feature identification samples to evaluate identification performance with severe teeth loss. In the feature extraction stage, we conduct voxel downsampling, then extract the geometric and structural features of the point cloud. In the registration-based identification stage, we construct a coarse-to-fine registration scheme in order to realize the identification task. Experimental results on a dataset of 160 individuals demonstrate that our method achieves a Rank-1 recognition rate of 100% using complete tooth crown contours samples. Utilizing the remaining four types of local feature samples yields a Rank-1 recognition rate exceeding 96.05%. The proposed framework proves effective for human identification, maintaining high identification performance even in extreme cases of partial tooth loss.

Letters, Words, Sentences, and Reading.

In this personal, and therefore highly selective, review article I summarize work performed in collaboration with numerous colleagues on how skilled adult readers perform identification tasks and speeded binary decision tasks involving single letters and visually presented words and sentences. The overarching aim is to highlight similarities in the processing performed at three key levels involved in written language comprehension (in languages that use an alphabetic script): letters, words, and sentences. The comparisons are made using behavioral data obtained with: i) speeded (response-limited) binary decision tasks; and ii) the effects of simultaneous surrounding context on letter and word identification using both data-limited (non-speeded) and response-limited procedures. I then propose a general framework that combines the three levels of processing, and that connects core processes at each level with the processing involved in tasks designed to reflect those core processes, and I end by suggesting possible avenues for future research with an aim to extend this general framework.

Modulation of audiovisual integration in the left and right sides: effects of side and spatial coherency

BackgroundUsing event-related potentials (ERPs), we aimed to investigate audiovisual integration neural mechanisms during a letter identification task in the left and right sides. Unimodal (A,V) and bimodal (AV) stimuli were presented on either side, with ERPs from unimodal (A,V) stimuli on the same side being compared to those from simultaneous bimodal stimuli (AV). Non-zero results of the AV-(A + V) difference waveforms indicated audiovisual integration on the left/right side.ResultsWhen spatially coherent AV stimuli were presented on the right side, two significant ERP components in the integrated differential wave were noted. The N134 and N262, present in the first 300 ms of the AV-(A + V) integration difference wave, indicated significant audiovisual integration effects. However, when these stimuli were presented on the left side, there were no significant integration components. This audiovisual integration difference may stem from left/right asymmetry of cerebral hemisphere language processing.ConclusionsAudiovisual letter information presented on the right side was easier to integrate, process, and represent. Additionally, only one significant integrative component peaked at 140 ms in the parietal cortex for spatially non-coherent AV stimuli and provided audiovisual multisensory integration, which could be attributed to some integrative neural processes that depend on the spatial congruity of the auditory and visual stimuli.

Tomato Leaf Disease Classification by Combining EfficientNetv2 and a Swin Transformer

Recently, convolutional neural networks (CNNs) and self-attention mechanisms have been widely applied in plant disease identification tasks, yielding significant successes. Currently, the majority of research models for tomato leaf disease recognition rely solely on traditional convolutional models or Transformer architectures and fail to capture both local and global features simultaneously. This limitation may result in biases in the model’s focus, consequently impacting the accuracy of disease recognition. Consequently, models capable of extracting local features while attending to global information have emerged as a novel research direction. To address these challenges, we propose an Eff-Swin model that integrates the enhanced features of the EfficientNetV2 and Swin Transformer networks, aiming to harness the local feature extraction capability of CNNs and the global modeling ability of Transformers. Comparative experiments demonstrate that the enhanced model has achieved a further increase in training accuracy, reaching an accuracy rate of 99.70% on the tomato leaf disease dataset, which is 0.49~3.68% higher than that of individual network models and 0.8~1.15% higher than that of existing state-of-the-art combined approaches. The results show that integrating attention mechanisms into convolutional models can significantly enhance the accuracy of tomato leaf disease recognition while also offering the great potential of the Eff-Swin backbone with self-attention in plant disease identification.

Parameter identification by deep learning of a material model for granular media

Classical physical modeling with associated numerical simulation (model-based), and prognostic methods based on the analysis of large amounts of data (data-driven) are the two most common methods used for the mapping of complex physical processes. In recent years, the efficient combination of these approaches has become increasingly important. Continuum mechanics in the core consists of conservation equations that-in addition to the always-necessary specification of the process conditions-can be supplemented by phenomenological material models. The latter are an idealized image of the specific material behavior that can be determined experimentally, empirically, and based on a wealth of expert knowledge. The more complex the material, the more difficult the calibration is. This situation forms the starting point for this work’s hybrid data-driven and model-based approach for mapping a complex physical process in continuum mechanics. Specifically, we use data generated from a classical physical model by the MESHFREE software (MESHFREE Team in Fraunhofer ITWM & SCAI: MESHFREE. https://www.meshfree.eu, 2023) to train a Principal Component Analysis-based neural network (PCA-NN) for the task of parameter identification of the material model parameters. The obtained results highlight the potential of deep-learning-based hybrid models for determining parameters, which are the key to characterizing materials occurring naturally such as sand, soil, mud, or snow. The motivation for our research is the simulation of the interaction of vehicles with sand. However, the applicability of the presented methodology is not limited to this industrial use case. In geosciences, when predicting the runout zones of landslides or avalanches and evaluating corresponding protective measures, the parameterization of the respective material model is essential.

Decision support improves information integration and criterion placement in a multi-cue signal-identification task

Identifying contacts in a military context can require operators to integrate multiple cues and to adjust response criteria to event base rates. The current experiment tested whether support from a decision aid would improve these processes. Participants performed a signal identification task that required them to integrate cues displayed as visual scale readings. In a static condition, participants saw a single set of readings each trial. In dynamic conditions, readings were updated over time. Base rates of signal categories were unequal, requiring participants to adopt biased response criteria to maximise response accuracy. Participants worked with or without an aid that combined cues and base rate information in an ideal manner. Support from the aid pushed participants’ response criteria towards optimal and improved integration of dynamic cues. Decision aids may be especially useful when task demands require biased response criteria and when cues are sampled over time.

Drive-By Identification of Joint Bridge Damage and Surface Roughness Based on Sensitivity Analysis of Residual Contact Point Deflections

Drive-by inspection of bridge damage using a passing vehicle’s dynamic response has great potential in bridge damage identification. Among the current drive-by methodologies, the methods based on bridge modeling have been proposed for the quantitative identification of bridge damage and bridge surface roughness. However, the coupling of bridge damage and unknown surface roughness increases the difficulty of the identification problem and most previous studies conduct the identification task of bridge damage or bridge surface roughness separately. In this paper, a novel method is proposed for the identification of joint bridge damage and bridge surface roughness based on the residual bridge deflections from the front and rear wheels contact points of an instrumented passing vehicle. The vehicle–bridge interaction is analyzed using a half-car model of a four degrees of freedom (4-DOF) with front and rear vehicle wheels. First, the vehicle–bridge unknown forces and displacement of the vehicle axles are identified by the generalized Kalman filter under unknown input (GKF-UI) proposed by the authors. The sensors can be installed conveniently on the vehicle body due to GKF-UI. Then, the residual bridge deflections from the front and rear vehicle wheels contact points are utilized to eliminate the effect of road surface roughness. Afterward, bridge structural damage is identified based on the sensitivity analysis of residual bridge deflections from the front and rear contact points with [Formula: see text]1-norm regularization. Finally, bridge road surface roughness is estimated based on the identified unknown forces and updated bridge model with damage. The results of numerical identification examples demonstrate the effectiveness of the proposed method for the identification of joint bridge damage and surface roughness.

Crystal structure identification with 3D convolutional neural networks with application to high-pressure phase transitions in SiO2

Efficient, reliable and easy-to-use structure recognition of atomic environments is essential for the analysis of atomic scale computer simulations. In this work, we train two neuronal network (NN) architectures, namely PointNet and dynamic graph convolutional NN (DG-CNN) using different hyperparameters and training regimes to assess their performance in structure identification tasks of atomistic structure data. We show benchmarks on simple crystal structures, where we can compare against established methods. The approach is subsequently extended to structurally more complex SiO2 phases. By making use of this structure recognition tool, we are able to achieve a deeper understanding of the crystallization process in amorphous SiO2 under shock compression. Lastly, we show how the NN based structure identification workflows can be integrated into OVITO using its python interface.

FedETC: Encrypted traffic classification based on federated learning

The current popular traffic classification methods based on feature engineering and machine learning are difficult to obtain suitable traffic feature sets for multiple traffic classification tasks. Besides, data privacy policies prohibit network operators from collecting and sharing traffic data that might compromise user privacy. To address these challenges, we propose FedETC, a federated learning framework that allows multiple participants to learn global traffic classifiers, while keeping locally encrypted traffic invisible to other participants. In addition, FedETC adopts one-dimensional convolutional neural network as the base model, which avoids manual traffic feature design. In the experiments, we evaluate the FedETC framework for the tasks of both application identification and traffic characterization in a publicly available real-world dataset. The results show that FedETC can achieve promising accuracy rates that are close to centralized learning schemes.

MODELS OF REMOTE IDENTIFICATION OF PARAMETERS OF DYNAMIC OBJECTS USING DETECTION TRANSFORMERS AND OPTICAL FLOW

The tasks of remote identification of parameters of dynamic objects are important for various fields, including computer vision, robotics, autonomous vehicles, video surveillance systems, and many others. Traditional methods of solving these problems face the problems of insufficient accuracy and efficiency of determining dynamic parameters in conditions of rapidly changing environments and complex dynamic scenarios. Modern methods of identifying parameters of dynamic objects using technologies of detection transformers and optical flow are considered. Transformer detection is one of the newest approaches in computer vision that uses transformer architecture for object detection tasks. This transformer integrates the object detection and boundary detection processes into a single end-to-end model, which greatly improves the accuracy and speed of processing. The use of transformers allows the model to effectively process information from the entire image at the same time, which contributes to better recognition of objects even in difficult conditions. Optical flow is a motion analysis method that determines the speed and direction of pixel movement between successive video frames. This method allows obtaining detailed information about the dynamics of the scene, which is critical for accurate tracking and identification of parameters of moving objects. The integration of detection transformers and optical flow is proposed to increase the accuracy of identification of parameters of dynamic objects. The combination of these two methods allows you to use the advantages of both approaches: high accuracy of object detection and detailed information about their movement. The conducted experiments show that the proposed model significantly outperforms traditional methods both in the accuracy of determining the parameters of objects and in the speed of data processing. The key results of the study indicate that the integration of detection transformers and optical flow provides reliable and fast determination of parameters of moving objects in real time, which can be applied in various practical scenarios. The conducted research also showed the potential for further improvement of data processing methods and their application in complex dynamic environments. The obtained results open new perspectives for the development of intelligent monitoring and control systems capable of adapting to rapidly changing environmental conditions, increasing the efficiency and safety of their work.

Contributions of Speaking, Listening, and Semantic Depth to Word Learning in Typical 3- and 4-Year-Olds.

This study examined learning via perception, learning via production, and semantic depth as contributors to word learning in preschool-aged children. There is broad evidence that semantic depth is an important contributor to word learning, especially when semantic cues are repeated and spaced out over time. Perceptual learning and production learning each support word learning sometimes, but not in all cases. The purpose of this study was to examine all three learning mechanisms within a single experimental paradigm. Thirty-six typically developing preschool children completed the experiment. They were familiarized with 16 novel words that were contextualized as alien names. These aliens came in four sets, each set comprising one base alien and three modified aliens marked by suffixes. Children completed four familiarizations: two in which they simply listened to the alien names (perceptual learning) and two where they repeated the alien names (production learning). Those conditions were crossed with a semantic depth manipulation (aliens with and without verbal semantic cues). Following each familiarization, referent identification and confrontation naming tasks were completed to assess learning. Children were able to identify more alien referents following familiarizations with semantic depth. There were no significant effects of either perceptual learning or production learning. This study confirms and expands on the benefits of semantic depth, but the results are unclear about the relative importance of perception and production to word learning. Nevertheless, the study suggests benefits to simultaneously studying multiple factors related to word learning.

AI-based rock strength assessment from tunnel face images using hybrid neural networks

In geological engineering and related fields, accurately and quickly identifying lithology and assessing rock strength are crucial for ensuring structural safety and optimizing design. Traditional rock strength assessment methods mainly rely on field sampling and laboratory tests, such as uniaxial compressive strength (UCS) tests and velocity tests. Although these methods provide relatively accurate rock strength data, they are complex, time-consuming, and unable to reflect real-time changes in field conditions. Therefore, this study proposes a new method based on artificial intelligence and neural networks to improve the efficiency and accuracy of rock strength assessments. This research utilizes a Transformer + UNet hybrid model for lithology identification and an optimized ResNet-18 model for determining rock weathering degrees, thereby correcting the strength of the tunnel face surrounding rock. Experimental results show that the Transformer + UNet hybrid model achieves an accuracy of 95.57% in lithology identification tasks, while the optimized ResNet model achieves an accuracy of 96.13% in rock weathering degree determination. Additionally, the average relative error in tunnel face strength detection results is only 9.33%, validating the feasibility and effectiveness of this method in practical engineering applications. The multi-model neural network system developed in this study significantly enhances prediction accuracy and efficiency, providing robust scientific decision support for tunnel construction, thereby improving construction safety and economy.

Особенности оценки вредоносной активности в инфраструктуре Умного города на основе гранулирования информации и гранулярных моделей вычислений

The object of the research is a new methodological approach to information granulation and fuzzy granular calculations, as a mathematical and methodological tool for improving the reliability of assessing the level of information security of the Smart City infrastructure. The proposed approach is one of the options for the practical application of elements of the theory of fuzzy sets in the tasks of search, identification and current assessment of signs of time-bearing activity. A detailed analysis of the features of this approach has been carried out, determining the expediency and conditions of its application for assessing malicious activity in the infrastructure of a Smart City. The theoretical aspects of the application of information granulation and fuzzy granular computing to the assessment of malicious activity combining various signs for various categories of potential threats to the infrastructure and subjects of a Smart City - the categories “cyberattack”, “malicious virus threat” or “data leakage (loss)” are studied and described. The analysis of the features of the proposed approach is carried out, which allows taking into account the opinions of experts and eliminating the vagueness associated with noise, disorder and lack of formalization of surveillance data collected and pre-processed in the interests of assessing threats and consequences of negative manifestations of malicious activity. A sequence of calculations and analytical expressions for calculating the estimated values of signs of harmful activity for various categories of potential threats to the infrastructure and subjects of a Smart City has been developed and described in detail. The approach assumes the practical possibility of evaluating signs of malicious activity using information granules formed on the basis of a minimum numerical distance between the values of membership functions characterizing vaguely specified data on the presence or absence of observed signs (attributes) of malicious activity, as well as granular summation and determination of the trace function of the granular sum. At the same time, the proposed approach makes it possible to obtain estimates of signs of malicious activity that are adequate to the tasks of monitoring the Smart City security policy and, ultimately, provides increased reliability of proactive threat control and analysis of the possible consequences of a negative manifestation of suspicious activity.

Handover for V2V communication in 5G using convolutional neural networks

Vehicle communication is one of the most vital aspects of modern transportation systems because it enables real-time data transmission between vehicles and infrastructure to improve traffic flow and road safety. The next generation of mobile technology, 5G, was created to address earlier generations' growing need for high data rates and quality of service issues. 5G cellular technology aims to eliminate penetration loss by segregating outside and inside settings and allowing extremely high transmission speeds, achieved by installing hundreds of dispersed antenna arrays using a distributed antenna system (DAS). Huge multiple-input multiple-output (MIMO) systems are accomplished via DASs and huge MIMO systems, where hundreds of dispersed antenna arrays are built. Because deep learning (DL) techniques employ artificial neural networks with at least one hidden layer, they are used in this study for vehicle recognition. They can swiftly process vast quantities of labeled training data to identify features. Therefore, this paper employed the VGG19 DL model through transfer learning to address the task of vehicle detection and obstacle identification. It also proposes a novel horizontal handover prediction method based on channel characteristics. The suggested techniques are designed for heterogeneous networks or horizontal handovers using DL. In the designated surrounding regions of 5G environments, the suggested detection and handover algorithms identified vehicles with a success rate of 97 % and predicted the next station for handover.

Identification, uncertainty quantification and validation of orthotropic material properties for additively manufactured polymers

The mechanical behavior of polymeric parts manufactured by fused filament fabrication is often characterized as orthotropic because of the layer-wise extrusion of the material. To perform reliable simulations of the mechanical response of these parts subjected to certain loads, the material parameters have to be known. This leads to the task of material parameter identification from suitable experimental data. In this work, the identification of the material parameters for orthotropic material behavior is studied in detail with a focus on the specific characteristics of additively manufactured parts. This comprises working out the differences that arise compared to the parameter identification for classical orthotropic composite materials. Moreover, the material parameters are identified from tensile and shear test information together with full-field displacement data from digital image correlation measurements. Then, the obtained parameters are compared for analytical and numerical approaches, where a non-linear least-squares method with finite elements is drawn on. The parameters are not only identified, but also an uncertainty analysis is carried out and, subsequently, the different methods are validated. Further, a clear derivation of commonly applied analytical equations is provided with a special focus on the underlying assumptions. It turns out that the analytical parameter identification procedures cannot be easily transferred to numerical procedures, especially when using full-field data, because of correlations between the parameters. Moreover, the identified parameters are employed to prove the reasonability of orthotropic material behavior for specimens manufactured by fused filament fabrication. Finally, during validation it turns out that all parameter identification methods provide a good agreement with experimental data, showing only small differences between the results of the different calibration strategies.

Two-stage transfer learning-based nonparametric system identification with Gaussian process regression

Most system identification methods ignore correlations between different identification tasks and do not make full use of historical models when identifying a new process. In this paper, a two-stage transfer learning-based nonparametric system identification method is proposed to improve model accuracy and to reduce identification costs. First, to overcome the transfer capability limitations of the discrete-time model, the target process is modeled as a continuous-time impulse response (IR) function, which consists of a transfer model part and a residual model part, which is regarded as a zero-mean Gaussian process (GP). Then, by utilizing the process geometrical characteristics, the IR functions are classified into three IR domains, and transformation functions within domain or between domains are designed to learn the relationship between the source process model and the target process model. Finally, a two-stage nonparametric identification algorithm based on GP regression is developed: The first stage is performed to select the appropriate type of transformation function through the weighted-derivative dynamical time warping technique, and the second stage is conducted to estimate the transfer model and residual model by using the empirical Bayes approach. Three case studies are conducted to validate the superiority of the proposed identification method.

A Multitask Learning-Based Vision Transformer for Plant Disease Localization and Classification

Plant disease detection is a critical task in agriculture, essential for ensuring crop health and productivity. Traditional methods in this context are often labor-intensive and prone to errors, highlighting the need for automated solutions. While computer vision-based solutions have been successfully deployed in recent years for plant disease identification and localization tasks, these often operate independently, leading to suboptimal performance. It is essential to develop an integrated solution combining these two tasks for improved efficiency and accuracy. This research proposes the innovative Plant Disease Localization and Classification model based on Vision Transformer (PDLC-ViT), which integrates co-scale, co-attention, and cross-attention mechanisms and a ViT, within a Multi-Task Learning (MTL) framework. The model was trained and evaluated on the Plant Village dataset. Key hyperparameters, including learning rate, batch size, dropout ratio, and regularization factor, were optimized through a thorough grid search. Early stopping based on validation loss was employed to prevent overfitting. The PDLC-ViT model demonstrated significant improvements in plant disease localization and classification tasks. The integration of co-scale, co-attention, and cross-attention mechanisms allowed the model to capture multi-scale dependencies and enhance feature learning, leading to superior performance compared to existing models. The PDLC-ViT model evaluated on two public datasets achieved an accuracy of 99.97%, a Mean Average Precision (MAP) of 99.18%, and a Mean Average Recall (MAR) of 99.11%. These results underscore the model's exceptional precision and recall, highlighting its robustness and reliability in detecting and classifying plant diseases. The PDLC-ViT model sets a new benchmark in plant disease detection, offering a reliable and advanced tool for agricultural applications. Its ability to integrate localization and classification tasks within an MTL framework promotes timely and accurate disease management, contributing to sustainable agriculture and food security.

Diabetic retinopathy identification based on multi-source-free domain adaptation.

To address the challenges of data labeling difficulties, data privacy, and necessary large amount of labeled data for deep learning methods in diabetic retinopathy (DR) identification, the aim of this study is to develop a source-free domain adaptation (SFDA) method for efficient and effective DR identification from unlabeled data. A multi-SFDA method was proposed for DR identification. This method integrates multiple source models, which are trained from the same source domain, to generate synthetic pseudo labels for the unlabeled target domain. Besides, a softmax-consistence minimization term is utilized to minimize the intra-class distances between the source and target domains and maximize the inter-class distances. Validation is performed using three color fundus photograph datasets (APTOS2019, DDR, and EyePACS). The proposed model was evaluated and provided promising results with respectively 0.8917 and 0.9795 F1-scores on referable and normal/abnormal DR identification tasks. It demonstrated effective DR identification through minimizing intra-class distances and maximizing inter-class distances between source and target domains. The multi-SFDA method provides an effective approach to overcome the challenges in DR identification. The method not only addresses difficulties in data labeling and privacy issues, but also reduces the need for large amounts of labeled data required by deep learning methods, making it a practical tool for early detection and preservation of vision in diabetic patients.