M. Yadollahzadeh Tabari; Z. Mataji
Abstract
The Internet of Things (IoT) is a novel paradigm in computer networks which is capable to connect things to the internet via a wide range of technologies. Due to the features of the sensors used in IoT networks and the unsecured nature of the internet, IoT is vulnerable to many internal routing attacks. ...
Read More
The Internet of Things (IoT) is a novel paradigm in computer networks which is capable to connect things to the internet via a wide range of technologies. Due to the features of the sensors used in IoT networks and the unsecured nature of the internet, IoT is vulnerable to many internal routing attacks. Using traditional IDS in these networks has its own challenges due to the resource constraint of the nodes, and the characteristics of the IoT network. A sinkhole attacker node, in this network, attempts to attract traffic through incorrect information advertisement. In this research, a distributed IDS architecture is proposed to detect sinkhole routing attack in RPL-based IoT networks, which is aimed to improve true detection rate and reduce the false alarms. For the latter we used one type of post processing mechanism in which a threshold is defined for separating suspicious alarms for further verifications. Also, the implemented IDS modules distributed via client and router border nodes that makes it energy efficient. The required data for interpretation of network’s behavior gathered from scenarios implemented in Cooja environment with the aim of Rapidminer for mining the produces patterns. The produced dataset optimized using Genetic algorithm by selecting appropriate features. We investigate three different classification algorithms which in its best case Decision Tree could reaches to 99.35 rate of accuracy.
M. Sepahvand; F. Abdali-Mohammadi
Abstract
The success of handwriting recognition methods based on digitizer-pen signal processing is mostly dependent on the defined features. Strong and discriminating feature descriptors can play the main role in improving the accuracy of pattern recognition. Moreover, most recognition studies utilize local ...
Read More
The success of handwriting recognition methods based on digitizer-pen signal processing is mostly dependent on the defined features. Strong and discriminating feature descriptors can play the main role in improving the accuracy of pattern recognition. Moreover, most recognition studies utilize local features or sequences of them. Whereas, it has been shown that the combination of global and local features can increase the recognition accuracy. This paper addresses two mentioned topics. First, a new high discriminative local feature, called Rotation Invariant Histogram of Degrees (RIHoD), is proposed for online digitizer-pen handwriting signals. Second, a feature representation layer is proposed, which maps local features into global ones in a new space using some learning kernels. Different aspects of the proposed local feature and learned global feature are analyzed and its efficiency is evaluated in several online handwriting recognition scenarios.
H.5. Image Processing and Computer Vision
Kimia Peyvandi
Abstract
Image inpainting is one of the important topics in the field of image processing, and various methods have been proposed in this area. However, this problem still faces multiple challenges, as an inpainting algorithm may perform well for a specific class of images but may have poor performance for other ...
Read More
Image inpainting is one of the important topics in the field of image processing, and various methods have been proposed in this area. However, this problem still faces multiple challenges, as an inpainting algorithm may perform well for a specific class of images but may have poor performance for other images. In this paper, we attempt to decompose the image into a low-rank component and a sparse component using (Principal Component Analysis) PCA, and then independently restore each component. For inpainting the low-rank component, we use an algorithm based on low-rank minimization, and for restoring the sparse component, we use the concept of splines. Using splines, we can effectively restore edges and lines, whereas the restoration of these regions is challenging in most algorithms. Also, in restoring the low-rank component, we construct a tensor at each step and approximate the missing pixels in the tensor, thereby significantly improving the efficiency of the low-rank minimization idea in image inpainting. Finally, we have applied our proposed method to restore various types of images, which demonstrates the effectiveness of our proposed method compared to other inpainting methods based on PSNR and SSIM.
M. Rahimi; A. A. Taheri; H. Mashayekhi
Abstract
Finding an effective way to combine the base learners is an essential part of constructing a heterogeneous ensemble of classifiers. In this paper, we propose a framework for heterogeneous ensembles, which investigates using an artificial neural network to learn a nonlinear combination of the base classifiers. ...
Read More
Finding an effective way to combine the base learners is an essential part of constructing a heterogeneous ensemble of classifiers. In this paper, we propose a framework for heterogeneous ensembles, which investigates using an artificial neural network to learn a nonlinear combination of the base classifiers. In the proposed framework, a set of heterogeneous classifiers are stacked to produce the first-level outputs. Then these outputs are augmented using several combination functions to construct the inputs of the second-level classifier. We conduct a set of extensive experiments on 121 datasets and compare the proposed method with other established and state-of-the-art heterogeneous methods. The results demonstrate that the proposed scheme outperforms many heterogeneous ensembles, and is superior compared to singly tuned classifiers. The proposed method is also compared to several homogeneous ensembles and performs notably better. Our findings suggest that the improvements are even more significant on larger datasets.
H.3.2.2. Computer vision
Masoumeh Esmaeiili; Kourosh Kiani
Abstract
The classification of emotions using electroencephalography (EEG) signals is inherently challenging due to the intricate nature of brain activity. Overcoming inconsistencies in EEG signals and establishing a universally applicable sentiment analysis model are essential objectives. This study introduces ...
Read More
The classification of emotions using electroencephalography (EEG) signals is inherently challenging due to the intricate nature of brain activity. Overcoming inconsistencies in EEG signals and establishing a universally applicable sentiment analysis model are essential objectives. This study introduces an innovative approach to cross-subject emotion recognition, employing a genetic algorithm (GA) to eliminate non-informative frames. Then, the optimal frames identified by the GA undergo spatial feature extraction using common spatial patterns (CSP) and the logarithm of variance. Subsequently, these features are input into a Transformer network to capture spatial-temporal features, and the emotion classification is executed using a fully connected (FC) layer with a Softmax activation function. Therefore, the innovations of this paper include using a limited number of channels for emotion classification without sacrificing accuracy, selecting optimal signal segments using the GA, and employing the Transformer network for high-accuracy and high-speed classification. The proposed method undergoes evaluation on two publicly accessible datasets, SEED and SEED-V, across two distinct scenarios. Notably, it attains mean accuracy rates of 99.96% and 99.51% in the cross-subject scenario, and 99.93% and 99.43% in the multi-subject scenario for the SEED and SEED-V datasets, respectively. Noteworthy is the outperformance of the proposed method over the state-of-the-art (SOTA) in both scenarios for both datasets, thus underscoring its superior efficacy. Additionally, comparing the accuracy of individual subjects with previous works in cross subject scenario further confirms the superiority of the proposed method for both datasets.
H.3.2.2. Computer vision
Maryam Baghi; Kourosh Kiani; Razieh Rastgoo
Abstract
With rapid advancements in information and communication technology, recommender systems have become vital tools across a wide range of online activities and e-commerce processes. Collaborative recommender systems, which utilize user data and contributions to provide suggestions, represent a significant ...
Read More
With rapid advancements in information and communication technology, recommender systems have become vital tools across a wide range of online activities and e-commerce processes. Collaborative recommender systems, which utilize user data and contributions to provide suggestions, represent a significant innovation in this field. In this paper, we conduct an analysis of collaborative recommender systems and evaluate their impact on enhancing the efficiency and accuracy of recommendations. To this end, we propose a deep learning approach using a Graph Convolutional Network (GCN), as a special type of Graph Neural Network (GNN). By assigning weights to edges between nodes, scores are calculated for these edges. The importance of the edges varies based on the number of neighboring nodes and their proximity to the target node. The higher the edge score, the more significant the path. To calculate edge weights, we leverage metrics such as Jaccard similarity, cosine similarity, LHN index, and Salton cosine similarity. This approach improves the identification of relationships between nodes and enhances the accuracy of the recommender system. For implementation, we utilized the well-known MovieLens dataset. Ultimately, users were clustered into 18 clusters, with a large number of nodes within each cluster. By clustering users, we increased the number and diversity of recommendations. This significantly improved the performance of the recommender system, yielding promising results.
H.3.8. Natural Language Processing
Milad Allhgholi; Hossein Rahmani; Amirhossein Derakhshan; Saman Mohammadi Raouf
Abstract
Document similarity matching is essential for efficient text retrieval, plagiarism detection, and content analysis. Existing studies in this field can be categorized into three approaches: statistical analysis, deep learning, and hybrid approaches. However, to the best of our knowledge, none have incorporated ...
Read More
Document similarity matching is essential for efficient text retrieval, plagiarism detection, and content analysis. Existing studies in this field can be categorized into three approaches: statistical analysis, deep learning, and hybrid approaches. However, to the best of our knowledge, none have incorporated the importance of named entities into their methodologies. In this paper, we propose DOSTE, a method that first extracts name entities and then utilizes them to enhance document similarity matching through statistical and graph-based analysis. Empirical results indicate that DOSTE achieves better results by emphasizing named entities, resulting in an average improvement of 9% in the average recall metric compared to baseline methods. Also, DOSTE unlike LLM-based approaches, does not require extensive GPU resources. Additionally, non-empirical interpretations of the results indicate that DOSTE is particularly effective in identifying similarity in short documents and complex document comparisons.
A. Lakizadeh; Z. Zinaty
Abstract
Aspect-level sentiment classification is an essential issue in sentiment analysis that intends to resolve the sentiment polarity of a specific aspect mentioned in the input text. Recent methods have discovered the role of aspects in sentiment polarity classification and developed various techniques to ...
Read More
Aspect-level sentiment classification is an essential issue in sentiment analysis that intends to resolve the sentiment polarity of a specific aspect mentioned in the input text. Recent methods have discovered the role of aspects in sentiment polarity classification and developed various techniques to assess the sentiment polarity of each aspect in the text. However, these studies do not pay enough attention to the need for vectors to be optimal for the aspect. To address this issue, in the present study, we suggest a Hierarchical Attention-based Method (HAM) for aspect-based polarity classification of the text. HAM works in a hierarchically manner; firstly, it extracts an embedding vector for aspects. Next, it employs these aspect vectors with information content to determine the sentiment of the text. The experimental findings on the SemEval2014 data set show that HAM can improve accuracy by up to 6.74% compared to the state-of-the-art methods in aspect-based sentiment classification task.
F. Baratzadeh; Seyed M. H. Hasheminejad
Abstract
With the advancement of technology, the daily use of bank credit cards has been increasing exponentially. Therefore, the fraudulent use of credit cards by others as one of the new crimes is also growing fast. For this reason, detecting and preventing these attacks has become an active area of study. ...
Read More
With the advancement of technology, the daily use of bank credit cards has been increasing exponentially. Therefore, the fraudulent use of credit cards by others as one of the new crimes is also growing fast. For this reason, detecting and preventing these attacks has become an active area of study. This article discusses the challenges of detecting fraudulent banking transactions and presents solutions based on deep learning. Transactions are examined and compared with other traditional models in fraud detection. According to the results obtained, optimal performance is related to the combined model of deep convolutional networks and short-term memory, which is trained using the aggregated data received from the generative adversarial network. This paper intends to produce sensible data to address the unequal class distribution problem, which is far more effective than traditional methods. Also, it uses the strengths of the two approaches by combining deep convolutional network and Long Short Term Memory network to improve performance. Due to the inefficiency of evaluation criteria such as accuracy in this application, the measure of distance score and the equal error rate has been used to evaluate models more transparent and more precise. Traditional methods were compared to the proposed approach to evaluate the efficiency of the experiment.
M. M. Jaziriyan; F. Ghaderi
Abstract
Most of the existing neural machine translation (NMT) methods translate sentences without considering the context. It is shown that exploiting inter and intra-sentential context can improve the NMT models and yield to better overall translation quality. However, providing document-level data is costly, ...
Read More
Most of the existing neural machine translation (NMT) methods translate sentences without considering the context. It is shown that exploiting inter and intra-sentential context can improve the NMT models and yield to better overall translation quality. However, providing document-level data is costly, so properly exploiting contextual data from monolingual corpora would help translation quality. In this paper, we proposed a new method for context-aware neural machine translation (CA-NMT) using a combination of hierarchical attention networks (HAN) and automatic post-editing (APE) techniques to fix discourse phenomena when there is lack of context. HAN is used when we have a few document-level data, and APE can be trained on vast monolingual document-level data to improve results further. Experimental results show that combining HAN and APE can complement each other to mitigate contextual translation errors and further improve CA-NMT by achieving reasonable improvement over HAN (i.e., BLEU score of 22.91 on En-De news-commentary dataset).
H.3.15.3. Evolutionary computing and genetic algorithms
Mahdieh Maazalahi; Soodeh Hosseini
Abstract
Detecting and preventing malware infections in systems is become a critical necessity. This paper presents a hybrid method for malware detection, utilizing data mining algorithms such as simulated annealing (SA), support vector machine (SVM), genetic algorithm (GA), and K-means. The proposed method combines ...
Read More
Detecting and preventing malware infections in systems is become a critical necessity. This paper presents a hybrid method for malware detection, utilizing data mining algorithms such as simulated annealing (SA), support vector machine (SVM), genetic algorithm (GA), and K-means. The proposed method combines these algorithms to achieve effective malware detection. Initially, the SA-SVM method is employed for feature selection, where the SVM algorithm identifies the best features, and the SA algorithm calculates the SVM parameters. Subsequently, the GA-K-means method is utilized to identify attacks. The GA algorithm selects the best chromosome for cluster centers, and the K-means algorithm has applied to identify malware. To evaluate the performance of the proposed method, two datasets, Andro-Autopsy and CICMalDroid 2020, have been utilized. The evaluation results demonstrate that the proposed method achieves high true positive rates (0.964, 0.985), true negative rates (0.985, 0.989), low false negative rates (0.036, 0.015), and false positive rates (0.022, 0.043). This indicates that the method effectively detects malware while reasonably minimizing false identifications.
H.3.2.2. Computer vision
Razieh Rastgoo
Abstract
Sign language (SL) is the primary mode of communication within the Deaf community. Recent advances in deep learning have led to the development of various applications and technologies aimed at facilitating bidirectional communication between the Deaf and hearing communities. However, challenges remain ...
Read More
Sign language (SL) is the primary mode of communication within the Deaf community. Recent advances in deep learning have led to the development of various applications and technologies aimed at facilitating bidirectional communication between the Deaf and hearing communities. However, challenges remain in the availability of suitable datasets for deep learning-based models. Only a few public large-scale annotated datasets are available for sign sentences, and none exist for Persian Sign Language sentences. To address this gap, we have collected a large-scale dataset comprising 10,000 sign sentence videos corresponding to 100 Persian sign sentences. This dataset includes comprehensive annotations such as the bounding box of the detected hand, class labels, hand pose parameters, and heatmaps. A notable feature of the proposed dataset is that it contains isolated signs corresponding to the sign sentences within the dataset. To analyze the complexity of the proposed dataset, we present extensive experiments and discuss the results. More concretely, the results of the models in key sub-domains relevant to Sign Language Recognition (SLR), including hand detection, pose estimation, real-time tracking, and gesture recognition, have been included and analyzed. Moreover, the results of seven deep learning-based models on the proposed datasets have been discussed. Finally, the results of Sign Language Production (SLP) using deep generative models have been presented. We report the experimental results of these models from these sub-areas, showcasing their performance on the proposed dataset.
H.3.15.3. Evolutionary computing and genetic algorithms
Homa Mehtarizadeh; Najme Mansouri; Behnam Mohammad Hasani Zade; Mohammad Mehdi Hosseini
Abstract
Accurate and reliable stock price prediction is both a formidable and essential task in financial markets, requiring the use of advanced techniques. This paper presents an innovative approach that integrates Long Short-Term Memory (LSTM) networks with Modified Complex Variational Mode Decomposition (MCVMD) ...
Read More
Accurate and reliable stock price prediction is both a formidable and essential task in financial markets, requiring the use of advanced techniques. This paper presents an innovative approach that integrates Long Short-Term Memory (LSTM) networks with Modified Complex Variational Mode Decomposition (MCVMD) for preprocessing and the Secretary Bird Optimization Algorithm (SBOA) for hyperparameter optimization. In the preprocessing phase, MCVMD decomposes stock price time series into intrinsic mode functions, effectively capturing complex patterns and reducing noise. To enhance predictive performance, SBOA optimizes both the hyperparameters of the LSTM network and the decomposition parameters of MCVMD. The proposed methodology is evaluated on datasets from six companies: Ferrari NV (RACE) and Intesa Sanpaolo (ISP) from Italy, Amadeus IT (AMA) and Repsol (REP) from Spain, and Hitachi Ltd (6501) and Chugai Pharmaceutical Co., Ltd. (4519) from Japan. Results show that the LSTM-MCVMD-SBOA model achieves lower error values compared with conventional benchmarks including ARIMA-GARCH, vanilla LSTM, Long Short-Term Memory-Particle Swarm Optimization (LSTM-PSO), and Long Short-Term Memory-Sine Cosine Algorithm (LSTM-SCA). Compared with these alternatives, SBOA was selected because of its superior balance between exploration and exploitation, inspired by secretary bird hunting and evasion behavior, which enables efficient search in complex optimization landscapes. Overall, the proposed model demonstrates significantly improved predictive accuracy over conventional methods, highlighting the efficacy of combining advanced decomposition with nature-inspired optimization for stock market forecasting.
A. Omondi; I. Lukandu; G. Wanyembi
Abstract
Variable environmental conditions and runtime phenomena require developers of complex business information systems to expose configuration parameters to system administrators. This allows system administrators to intervene by tuning the bottleneck configuration parameters in response to current changes ...
Read More
Variable environmental conditions and runtime phenomena require developers of complex business information systems to expose configuration parameters to system administrators. This allows system administrators to intervene by tuning the bottleneck configuration parameters in response to current changes or in anticipation of future changes in order to maintain the system’s performance at an optimum level. However, these manual performance tuning interventions are prone to error and lack of standards due to fatigue, varying levels of expertise and over-reliance on inaccurate predictions of future states of a business information system. As a result, the purpose of this research is to investigate on how the capacity of probabilistic reasoning to handle uncertainty can be combined with the capacity of Markov chains to map stochastic environmental phenomena to ideal self-optimization actions. This was done using a comparative experimental research design that involved quantitative data collection through simulations of different algorithm variants. This provided compelling results that indicate that applying the algorithm in a distributed database system improves performance of tuning decisions under uncertainty. The improvement was quantitatively measured by a response-time latency that was 27% lower than average and a transaction throughput that was 17% higher than average.
H.R. Koosha; Z. Ghorbani; R. Nikfetrat
Abstract
In the last decade, online shopping has played a vital role in customers' approach to purchasing different products, providing convenience to shop and many benefits for the economy. E-commerce is widely used for digital media products such as movies, images, and software. So, recommendation systems are ...
Read More
In the last decade, online shopping has played a vital role in customers' approach to purchasing different products, providing convenience to shop and many benefits for the economy. E-commerce is widely used for digital media products such as movies, images, and software. So, recommendation systems are of great importance, especially in today's hectic world, which search for content that would be interesting to an individual. In this research, a new two-steps recommender system is proposed based on demographic data and user ratings on the public MovieLens datasets. In the first step, clustering on the training dataset is performed based on demographic data, grouping customers in homogeneous clusters. The clustering includes a hybrid Firefly Algorithm (FA) and K-means approach. Due to the FA's ability to avoid trapping into local optima, which resolves K-means' main pitfall, the combination of these two techniques leads to much better performance. In the next step, for each cluster, two recommender systems are proposed based on K-Nearest Neighbor (KNN) and Naïve Bayesian Classification. The results are evaluated based on many internal and external measures like the Davies-Bouldin index, precision, accuracy, recall, and F-measure. The results showed the effectiveness of the K-means/FA/KNN compared with other extant models.
H.3. Artificial Intelligence
M. Taghian; A. Asadi; R. Safabakhsh
Abstract
The quality of the extracted features from a long-term sequence of raw prices of the instruments greatly affects the performance of the trading rules learned by machine learning models. Employing a neural encoder-decoder structure to extract informative features from complex input time-series has proved ...
Read More
The quality of the extracted features from a long-term sequence of raw prices of the instruments greatly affects the performance of the trading rules learned by machine learning models. Employing a neural encoder-decoder structure to extract informative features from complex input time-series has proved very effective in other popular tasks like neural machine translation and video captioning. In this paper, a novel end-to-end model based on the neural encoder-decoder framework combined with deep reinforcement learning is proposed to learn single instrument trading strategies from a long sequence of raw prices of the instrument. In addition, the effects of different structures for the encoder and various forms of the input sequences on the performance of the learned strategies are investigated. Experimental results showed that the proposed model outperforms other state-of-the-art models in highly dynamic environments.
H.5. Image Processing and Computer Vision
Sekine Asadi Amiri; Mahda Nasrolahzadeh; Zeynab Mohammadpoory; AbdolAli Movahedinia; Amirhossein Zare
Abstract
Improving the quality of food industries and the safety and health of the people’s nutrition system is one of the important goals of governments. Fish is an excellent source of protein. Freshness is one of the most important quality criteria for fish that should be selected for consumption. It ...
Read More
Improving the quality of food industries and the safety and health of the people’s nutrition system is one of the important goals of governments. Fish is an excellent source of protein. Freshness is one of the most important quality criteria for fish that should be selected for consumption. It has been shown that due to improper storage conditions of fish, bacteria, and toxins may cause diseases for human health. The conventional methods of detecting spoilage and disease in fish, i.e. analyzing fish samples in the laboratory, are laborious and time-consuming. In this paper, an automatic method for identifying spoiled fish from fresh fish is proposed. In the proposed method, images of fish eyes are used. Fresh fish are identified by shiny eyes, and poor and stale fish are identified by gray color changes in the eye. In the proposed method, Inception-ResNet-v2 convolutional neural network is used to extract features. To increase the accuracy of the model and prevent overfitting, only some useful features are selected using the mRMR feature selection method. The mRMR reduces the dimensionality of the data and improves the classification accuracy. Then, since the number of samples is low, the k-fold cross-validation method is used. Finally, for classifying the samples, Naïve bayes and Random forest classifiers are used. The proposed method has reached an accuracy of 97% on the fish eye dataset, which is better than previous references.
H.3.8. Natural Language Processing
Ali Reza Ghasemi; Javad Salimi Sartakhti
Abstract
This paper evaluates the performance of various fine-tuning methods in Persian natural language processing (NLP) tasks. In low-resource languages like Persian, which suffer from a lack of rich and sufficient data for training large models, it is crucial to select appropriate fine-tuning ...
Read More
This paper evaluates the performance of various fine-tuning methods in Persian natural language processing (NLP) tasks. In low-resource languages like Persian, which suffer from a lack of rich and sufficient data for training large models, it is crucial to select appropriate fine-tuning techniques that mitigate overfitting and prevent the model from learning weak or surface-level patterns. The main goal of this research is to compare the effectiveness of fine-tuning approaches such as Full-Finetune, LoRA, AdaLoRA, and DoRA on model learning and task performance. We apply these techniques to three different Persian NLP tasks: sentiment analysis, named entity recognition (NER), and span question answering (QA). For this purpose, we conduct experiments on three Transformer-based multilingual models with different architectures and parameter scales: BERT-base multilingual (~168M parameters) with Encoder only structure, mT5-small (~300M parameters) with Encoder-Decoder structure, and mGPT (~1.4B parameters) with Decoder only structure. Each of these models supports the Persian language but varies in structure and computational requirements, influencing the effectiveness of different fine-tuning approaches. Results indicate that fully fine-tuned BERT-base multilingual consistently outperforms other models across all tasks in basic metrics, particularly given the unique challenges of these embedding-based tasks. Additionally, lightweight fine-tuning methods like LoRA and DoRA offer very competitive performance while significantly reducing computational overhead and outperform other models in Performance-Efficiency Score introduced in the paper. This study contributes to a better understanding of fine-tuning methods, especially for Persian NLP, and offers practical guidance for applying Large Language Models (LLMs) to downstream tasks in low-resource languages.
M. Zarbazoo Siahkali; A.A. Ghaderi; Abdol H. Bahrpeyma; M. Rashki; N. Safaeian Hamzehkolaei
Abstract
Scouring, occurring when the water flow erodes the bed materials around the bridge pier structure, is a serious safety assessment problem for which there are many equations and models in the literature to estimate the approximate scour depth. This research is aimed to study how surrogate models estimate ...
Read More
Scouring, occurring when the water flow erodes the bed materials around the bridge pier structure, is a serious safety assessment problem for which there are many equations and models in the literature to estimate the approximate scour depth. This research is aimed to study how surrogate models estimate the scour depth around circular piers and compare the results with those of the empirical formulations. To this end, the pier scour depth was estimated in non-cohesive soils based on a subcritical flow and live bed conditions using the artificial neural networks (ANN), group method of data handling (GMDH), multivariate adaptive regression splines (MARS) and Gaussian process models (Kriging). A database containing 246 lab data gathered from various studies was formed and the data were divided into three random parts: 1) training, 2) validation and 3) testing to build the surrogate models. The statistical error criteria such as the coefficient of determination (R2), root mean squared error (RMSE), mean absolute percentage error (MAPE) and absolute maximum percentage error (MPE) of the surrogate models were then found and compared with those of the popular empirical formulations. Results revealed that the surrogate models’ test data estimations were more accurate than those of the empirical equations; Kriging has had better estimations than other models. In addition, sensitivity analyses of all surrogate models showed that the pier width’s dimensionless expression (b/y) had a greater effect on estimating the normalized scour depth (Ds/y).
H.3.7. Learning
Mohammad Rezaei; Mohsen Rezvani; Morteza Zahedi
Abstract
With the increasing interconnectedness of communications and social networks, graph-based learning techniques offer valuable information extraction from data. Traditional centralized learning methods faced challenges, including data privacy violations and costly maintenance in a centralized environment. ...
Read More
With the increasing interconnectedness of communications and social networks, graph-based learning techniques offer valuable information extraction from data. Traditional centralized learning methods faced challenges, including data privacy violations and costly maintenance in a centralized environment. To address these, decentralized learning approaches like Federated Learning have emerged. This study explores the significant attention Federated Learning has gained in graph classification and investigates how Model Agnostic Meta-Learning (MAML) can improve its performance, especially concerning non-IID (Non-Independent Identically Distributed) data distributions.In real-world scenarios, deploying Federated Learning poses challenges, particularly in tuning client parameters and structures due to data isolation and diversity. To address this issue, this study proposes an innovative approach using Genetic Algorithms (GA) for automatic tuning of structures and parameters. By integrating GA with MAML-based clients in Federated Learning, various aspects, such as graph classification structure, learning rate, and optimization function type, can be automatically adjusted. This novel approach yields improved accuracy in decentralized learning at both the client and server levels.
H.3.8. Natural Language Processing
Arash Keshtkar; Saeedeh Sadat Sadidpour; Hossien Shirazi
Abstract
Word Sense Disambiguation (WSD) is a longstanding challenge in natural language processing, particularly in morphologically rich and low-resource languages such as Persian. The inherent ambiguity of Persian named entities exacerbated by domain-specific contexts and limited labeled data complicates both ...
Read More
Word Sense Disambiguation (WSD) is a longstanding challenge in natural language processing, particularly in morphologically rich and low-resource languages such as Persian. The inherent ambiguity of Persian named entities exacerbated by domain-specific contexts and limited labeled data complicates both semantic interpretation and information extraction. In this study, we introduce the PWNC corpus, a large-scale, integrated dataset designed for both Named Entity Recognition (NER) and WSD in Persian. The corpus was automatically constructed through a semi-supervised framework, incorporating contextual similarity measures and clustering algorithms to annotate ambiguous entities across ten semantic categories. Utilizing a semi-supervised framework, the proposed homograph semantic categorization method achieved robust performance, with a precision of 83%, recall of 81%, and an F1-score of 82% across over 305K annotated paragraphs. Detailed error analysis revealed challenges in disambiguating closely related senses and weak entities, which were mitigated through contextual embedding strategies. This work provides the first publicly available dual-task corpus for Persian NER and WSD, offering a scalable solution for disambiguation in low-resource tasks and laying the baseline for future research in Persian semantic processing.
M. Khanzadi; H. Veisi; R. Alinaghizade; Z. Soleymani
Abstract
One of the main problems in children with learning difficulties is the weakness of phonological awareness (PA) skills. In this regard, PA tests are used to evaluate this skill. Currently, this assessment is paper-based for the Persian language. To accelerate the process of the assessments and make it ...
Read More
One of the main problems in children with learning difficulties is the weakness of phonological awareness (PA) skills. In this regard, PA tests are used to evaluate this skill. Currently, this assessment is paper-based for the Persian language. To accelerate the process of the assessments and make it engaging for children, we propose a computer-based solution that is a comprehensive Persian phonological awareness assessment system implementing expressive and pointing tasks. For the expressive tasks, the solution is powered by recurrent neural network-based speech recognition systems. To this end, various recognition modules are implemented, including a phoneme recognition system for the phoneme segmentation task, a syllable recognition system for the syllable segmentation task, and a sub-word recognition system for three types of phoneme deletion tasks, including initial, middle, and final phoneme deletion. The recognition systems use bidirectional long short-term memory neural networks to construct acoustic models. To implement the recognition systems, we designed and collected Persian Kid’s Speech Corpus that is the largest in Persian for children’s speech. The accuracy rate for phoneme recognition was 85.5%, and for syllable recognition was 89.4%. The accuracy rates of the initial, middle, and final phoneme deletion were 96.76%, 98.21%, and 95.9%, respectively.
P. Abdzadeh; H. Veisi
Abstract
Automatic Speaker Verification (ASV) systems have proven to bevulnerable to various types of presentation attacks, among whichLogical Access attacks are manufactured using voiceconversion and text-to-speech methods. In recent years, there has beenloads of work concentrating on synthetic speech detection, ...
Read More
Automatic Speaker Verification (ASV) systems have proven to bevulnerable to various types of presentation attacks, among whichLogical Access attacks are manufactured using voiceconversion and text-to-speech methods. In recent years, there has beenloads of work concentrating on synthetic speech detection, and with the arrival of deep learning-based methods and their success in various computer science fields, they have been a prevailing tool for this very task too. Most of the deep neural network-based techniques forsynthetic speech detection have employed the acoustic features basedon Short-Term Fourier Transform (STFT), which are extracted from theraw audio signal. However, lately, it has been discovered that the usageof Constant Q Transform's (CQT) spectrogram can be a beneficialasset both for performance improvement and processing power andtime reduction of a deep learning-based synthetic speech detection. In this work, we compare the usage of the CQT spectrogram and some most utilized STFT-based acoustic features. As lateral objectives, we consider improving the model's performance as much as we can using methods such as self-attention and one-class learning. Also, short-duration synthetic speech detection has been one of the lateral goals too. Finally, we see that the CQT spectrogram-based model not only outperforms the STFT-based acoustic feature extraction methods but also reduces the processing time and resources for detecting genuine speech from fake. Also, the CQT spectrogram-based model places wellamong the best works done on the LA subset of the ASVspoof 2019 dataset, especially in terms of Equal Error Rate.
H.5. Image Processing and Computer Vision
Khosro Rezaee
Abstract
Bipolar disorder (BD) remains a pervasive mental health challenge, demanding innovative diagnostic approaches beyond traditional, subjective assessments. This study pioneers a non-invasive handwriting-based diagnostic framework, leveraging the unique interplay between psychological states and motor expressions ...
Read More
Bipolar disorder (BD) remains a pervasive mental health challenge, demanding innovative diagnostic approaches beyond traditional, subjective assessments. This study pioneers a non-invasive handwriting-based diagnostic framework, leveraging the unique interplay between psychological states and motor expressions in writing. Our hybrid deep learning model, combining ResNet for intricate feature extraction and external attention mechanisms for global pattern analysis, achieves a remarkably high accuracy 99%, validated through Leave-One-Subject-Out (LOSO) cross-validation. Augmented with advanced data preprocessing and augmentation techniques, the framework adeptly addresses dataset imbalances and handwriting variability. For the first time, Persian handwriting serves as a medium, bridging cultural gaps in BD diagnostics. This work not only establishes handwriting as a transformative tool for mental health diagnostics but also sets the stage for accessible, scalable, and culturally adaptive solutions in global mental healthcare.
R. Ghotboddini; H. Toossian Shandiz
Abstract
Lighting continuity is one of the preferences of citizens. Public lighting management from the viewpoint of city residents improves social welfare. The quality of lamps and duration of lighting defect correction is important in lighting continuity. In this regard, reward and penalty mechanism plays an ...
Read More
Lighting continuity is one of the preferences of citizens. Public lighting management from the viewpoint of city residents improves social welfare. The quality of lamps and duration of lighting defect correction is important in lighting continuity. In this regard, reward and penalty mechanism plays an important role in contract. Selecting labor and lamps has a significant impact on risk reduction during the contract period. This research improves strategies for public lighting asset management. The lifespan of lamp that announced by manufacturers is used to calculate maintenance cost in order to provide a possibility to estimate the actual cost of high-pressure sodium luminaire in public lighting system. Guarantee period of lamps and maximum permissible lighting defect detection and correction time is used for reward and penalty mechanism. The result shows that the natural life guarantee and permissible correction time have a considerable effect in maintenance cost and city resident’s satisfaction.
