Seyed Abolghasem Mirroshandel
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Seyed Abolghasem Mirroshandel has a total of 91 co-authors. Here are the top co-authors by number of shared publications:
- Arash Yousefi Jordehi
University of Guilan
4 shared publications Fatemeh Ghasemian
Department of Biology, University of Guilan, Rasht, Iran
4 shared publications- Arash Yousefi Jordehi
University of Guilan
2 shared publications - Cornelia Caragea
Computer Science University of Illinois Chicago
2 shared publications - Erfan Miahi
Independent AI Researcher
2 shared publications Gholamreza Ghassem-Sani
Computer Engineering Department Sharif University of Technology Tehran Iran
2 shared publications
Latest Publications
- Open Access
- 18 September 2026
- Published by Elsevier in Data in Brief
- Vol. 69
A multilingual annotated dataset for metadata extraction from Persian, Arabic and English scientific articles: header and reference section annotations
- 18 September 2026
- Published by Elsevier in Data in Brief
- Vol. 69
Open AccessA multilingual annotated dataset for metadata extraction from Persian, Arabic and English scientific articles: header and reference section annotations
- 1 September 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (17) , 175232
Seq2Seq2Seq: lossless data compression via discrete latent transformers and reinforcement learning
Efficient lossless compression is essential for reducing storage and transmission requirements while preserving data exactly. Traditional dictionary-based and statistical compressors can be limited in their ability to exploit complex, long-range dependencies. In this paper, we propose a lossless compression framework that combines a T5-small architecture with Advantage Actor-Critic reinforcement learning (RL) to generate a variable-length sequence of discrete compressed tokens. Rather than relying on a continuous autoencoder bottleneck, the proposed framework directly optimizes the length of this discrete representation while preserving the information required for exact reconstruction. The method operates without external grammatical rules or world knowledge. On the enwik8 benchmark, it achieves a compression ratio of 4.14, improving upon XZ (4.05) by approximately 2.3% and GZIP (2.74) by 51.0%. Although this result remains below neural network compression v3.2 (6.70), it demonstrates that RL can learn a compact, lossless discrete representation using two T5-small networks evaluated on a single 12 GB GPU device.
- 1 September 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (17) , 175232
Seq2Seq2Seq: lossless data compression via discrete latent transformers and reinforcement learning
Efficient lossless compression is essential for reducing storage and transmission requirements while preserving data exactly. Traditional dictionary-based and statistical compressors can be limited in their ability to exploit complex, long-range dependencies. In this paper, we propose a lossless compression framework that combines a T5-small architecture with Advantage Actor-Critic reinforcement learning (RL) to generate a variable-length sequence of discrete compressed tokens. Rather than relying on a continuous autoencoder bottleneck, the proposed framework directly optimizes the length of this discrete representation while preserving the information required for exact reconstruction. The method operates without external grammatical rules or world knowledge. On the enwik8 benchmark, it achieves a compression ratio of 4.14, improving upon XZ (4.05) by approximately 2.3% and GZIP (2.74) by 51.0%. Although this result remains below neural network compression v3.2 (6.70), it demonstrates that RL can learn a compact, lossless discrete representation using two T5-small networks evaluated on a single 12 GB GPU device.
- Open Access
- 9 July 2026
- Published by Elsevier in Informatics in Medicine Unlocked
- Vol. 64
Deep learning workflow for 3D calcification segmentation in cone beam computed tomographic images
- 9 July 2026
- Published by Elsevier in Informatics in Medicine Unlocked
- Vol. 64
Open AccessDeep learning workflow for 3D calcification segmentation in cone beam computed tomographic images
- 1 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Towards less supervision in dependency parsing
Analyse probabiliste est l'un des domaines de recherche les plus attractives en langage naturel En traitement. Analyseurs probabilistes succès actuels nécessitent de grandes treebanks qui Il est difficile, prend du temps et coûteux à produire. Par conséquent, nous avons concentré notre l'attention sur des approches moins supervisés. Nous avons proposé deux catégories de solution: l'apprentissage actif et l'algorithme semi-supervisé. Stratégies d'apprentissage actives permettent de sélectionner les échantillons les plus informatives pour annotation. La plupart des stratégies d'apprentissage actives existantes pour l'analyse reposent sur la sélection phrases incertaines pour l'annotation. Nous montrons dans notre recherche, sur quatre différents langues (français, anglais, persan, arabe), que la sélection des phrases complètes ne sont pas une solution optimale et de proposer un moyen de sélectionner uniquement les sous-parties de phrases. Comme nos expériences ont montré, certaines parties des phrases ne contiennent aucune utiles information pour la formation d'un analyseur, et en se concentrant sur les sous-parties incertains des phrases est une solution plus efficace dans l'apprentissage actif.
- 1 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Towards less supervision in dependency parsing
Analyse probabiliste est l'un des domaines de recherche les plus attractives en langage naturel En traitement. Analyseurs probabilistes succès actuels nécessitent de grandes treebanks qui Il est difficile, prend du temps et coûteux à produire. Par conséquent, nous avons concentré notre l'attention sur des approches moins supervisés. Nous avons proposé deux catégories de solution: l'apprentissage actif et l'algorithme semi-supervisé. Stratégies d'apprentissage actives permettent de sélectionner les échantillons les plus informatives pour annotation. La plupart des stratégies d'apprentissage actives existantes pour l'analyse reposent sur la sélection phrases incertaines pour l'annotation. Nous montrons dans notre recherche, sur quatre différents langues (français, anglais, persan, arabe), que la sélection des phrases complètes ne sont pas une solution optimale et de proposer un moyen de sélectionner uniquement les sous-parties de phrases. Comme nos expériences ont montré, certaines parties des phrases ne contiennent aucune utiles information pour la formation d'un analyseur, et en se concentrant sur les sous-parties incertains des phrases est une solution plus efficace dans l'apprentissage actif.
- 27 February 2026
- Published by AI Access Foundation in Journal of Artificial Intelligence Research
- Vol. 85
Label-Aware Pseudo-Training Sample Generation for Text Classification
Deep learning models excel in various Natural Language Processing (NLP) tasks, but their performance (excluding approaches like zero-shot learning or few-shot learning) relies on ample data, posing challenges in fields with limited datasets. To address the poverty in the size of training data, a number of approaches could be taken, such as multi-task learning and data augmentation. Aiming to leverage Large Language Models (LLMs), we propose a data augmentation algorithm. It subtly alters sentences by inserting random words and utilizes LLMs to find the most fitting replacements within their embedding space. Taking inspiration from Prompt Tuning, the focus shifts from optimizing the input prompt to updating the inserted tokens’ embedding vectors by maximizing the conditional generation probability. This allows for vast sample generation while implicitly benefiting from the knowledge within LLMs. The results from our extensive set of experiments on various benchmark text classification tasks show a substantial improvement over the non-augmented outcomes.
- 27 February 2026
- Published by AI Access Foundation in Journal of Artificial Intelligence Research
- Vol. 85
Label-Aware Pseudo-Training Sample Generation for Text Classification
Deep learning models excel in various Natural Language Processing (NLP) tasks, but their performance (excluding approaches like zero-shot learning or few-shot learning) relies on ample data, posing challenges in fields with limited datasets. To address the poverty in the size of training data, a number of approaches could be taken, such as multi-task learning and data augmentation. Aiming to leverage Large Language Models (LLMs), we propose a data augmentation algorithm. It subtly alters sentences by inserting random words and utilizes LLMs to find the most fitting replacements within their embedding space. Taking inspiration from Prompt Tuning, the focus shifts from optimizing the input prompt to updating the inserted tokens’ embedding vectors by maximizing the conditional generation probability. This allows for vast sample generation while implicitly benefiting from the knowledge within LLMs. The results from our extensive set of experiments on various benchmark text classification tasks show a substantial improvement over the non-augmented outcomes.
- Top Cited PapersOpen Access
- 25 February 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
A novel lightweight hybrid CNN–ViT for maize leaf disease classification
Maize is a vital global crop, but its productivity is often threatened by plant diseases, highlighting the need for precise and timely diagnostic methods. Traditional manual inspection is inefficient and prone to errors, motivating the development of automated solutions. Recent advances in computer vision and deep learning have enabled effective automated plant disease diagnosis. While Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) have shown promise in plant disease classification, CNNs struggle to capture global contextual information, and ViTs require large datasets and high computational resources. Inspired by mixture-of-experts (MoE) architectures, we propose a lightweight hybrid model that integrates CNN and ViT components, adaptively emphasizing local or global features based on input characteristics. Evaluated on a novel, real-world dataset of full maize plant images, our approach achieves 99.90% classification accuracy, significantly outperforming state-of-the-art baselines such as MobileViT, PiT, EdgeNeXt, and DeiT. These results demonstrate that lightweight hybrid architectures can deliver high-performance disease diagnosis suitable for practical agricultural deployment. The code is available at: https://www.github.com/sabermehdipour/MXiT.
- 25 February 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Top Cited PapersOpen AccessA novel lightweight hybrid CNN–ViT for maize leaf disease classification
Maize is a vital global crop, but its productivity is often threatened by plant diseases, highlighting the need for precise and timely diagnostic methods. Traditional manual inspection is inefficient and prone to errors, motivating the development of automated solutions. Recent advances in computer vision and deep learning have enabled effective automated plant disease diagnosis. While Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) have shown promise in plant disease classification, CNNs struggle to capture global contextual information, and ViTs require large datasets and high computational resources. Inspired by mixture-of-experts (MoE) architectures, we propose a lightweight hybrid model that integrates CNN and ViT components, adaptively emphasizing local or global features based on input characteristics. Evaluated on a novel, real-world dataset of full maize plant images, our approach achieves 99.90% classification accuracy, significantly outperforming state-of-the-art baselines such as MobileViT, PiT, EdgeNeXt, and DeiT. These results demonstrate that lightweight hybrid architectures can deliver high-performance disease diagnosis suitable for practical agricultural deployment. The code is available at: https://www.github.com/sabermehdipour/MXiT.
- Open Access
Seq2Seq2Seq: Lossless Data Compression via Discrete Latent Transformers and Reinforcement Learning
Efficient lossless compression is essential for minimizing storage costs and transmission overhead while preserving data integrity. Traditional compression techniques, such as dictionary-based and statistical methods, often struggle to optimally exploit the structure and redundancy in complex data formats. Recent advancements in deep learning have opened new avenues for compression; however, many existing approaches depend on dense vector representations that obscure the underlying token structure. To address these limitations, we propose a novel lossless compression method that leverages Reinforcement Learning applied to a T5 language model architecture. This approach enables the compression of data into sequences of tokens rather than traditional vector representations. Unlike auto-encoders, which typically encode information into continuous latent spaces, our method preserves the token-based structure, aligning more closely with the original data format. This preservation allows for higher compression ratios while maintaining semantic integrity. By training the model using an off-policy Reinforcement Learning algorithm, we optimize sequence length to minimize redundancy and enhance compression efficiency. Our method introduces an efficient and adaptive data compression system built upon advanced Reinforcement Learning techniques, functioning independently of external grammatical or world knowledge. This approach shows significant improvements in compression ratios compared to conventional methods. By leveraging the latent information within language models, our system effectively compresses data without requiring explicit content understanding, paving the way for more robust and practical compression solutions across various applications.
Open AccessSeq2Seq2Seq: Lossless Data Compression via Discrete Latent Transformers and Reinforcement Learning
Efficient lossless compression is essential for minimizing storage costs and transmission overhead while preserving data integrity. Traditional compression techniques, such as dictionary-based and statistical methods, often struggle to optimally exploit the structure and redundancy in complex data formats. Recent advancements in deep learning have opened new avenues for compression; however, many existing approaches depend on dense vector representations that obscure the underlying token structure. To address these limitations, we propose a novel lossless compression method that leverages Reinforcement Learning applied to a T5 language model architecture. This approach enables the compression of data into sequences of tokens rather than traditional vector representations. Unlike auto-encoders, which typically encode information into continuous latent spaces, our method preserves the token-based structure, aligning more closely with the original data format. This preservation allows for higher compression ratios while maintaining semantic integrity. By training the model using an off-policy Reinforcement Learning algorithm, we optimize sequence length to minimize redundancy and enhance compression efficiency. Our method introduces an efficient and adaptive data compression system built upon advanced Reinforcement Learning techniques, functioning independently of external grammatical or world knowledge. This approach shows significant improvements in compression ratios compared to conventional methods. By leveraging the latent information within language models, our system effectively compresses data without requiring explicit content understanding, paving the way for more robust and practical compression solutions across various applications.
- Open Access
- 7 February 2026
- Published by Elsevier in Intelligence-Based Medicine
- Vol. 13
MoE-Net: A deep ensemble framework optimized by genetic algorithm for nasal bone fracture detection on lateral X-ray images
- 7 February 2026
- Published by Elsevier in Intelligence-Based Medicine
- Vol. 13
Open AccessMoE-Net: A deep ensemble framework optimized by genetic algorithm for nasal bone fracture detection on lateral X-ray images
- Open Access
- 7 January 2026
- Published by Ubiquity Press, Ltd. in Transactions of the International Society for Music Information Retrieval
- Vol. 9 (1) , 1-20
Supervised Contrastive Models for Music Information Retrieval in Classical Persian Music
Musical instrument classification is a key task in music information retrieval, supporting applications such as automatic transcription and music recommendation. Research on this topic for traditional Persian music has been limited, largely due to the lack of complete and consistent datasets. This study introduces a reproducible baseline framework that combines supervised contrastive learning with stacked slice‑level aggregation—a late‑fusion approach that integrates predictions from short one‑second segments—to classify 15 traditional Persian instruments, including Ney, Setar, Tar, Santur, and Kamancheh. To support this work, we present the Persian Classical Instrument Dataset, a curated collection of publicly available solo‑instrument recordings encompassing 15 classical Persian instruments. Experiments across three settings—the five‑instrument subset, the full 15‑instrument dataset, and an existing baseline dataset—show that the proposed method achieves up to 99% accuracy on smaller subsets and 98% accuracy on the full dataset using 30‑second inputs. Furthermore, the same framework demonstrates strong generalization on the Dastgah detection task, outperforming previous methods and indicating that timbre‑based representations can transfer effectively to higher‑level modal recognition. Overall, this work provides both a publicly available dataset and a transparent, general‑purpose baseline to advance research on Persian and other non‑Western musical traditions.
- 7 January 2026
- Published by Ubiquity Press, Ltd. in Transactions of the International Society for Music Information Retrieval
- Vol. 9 (1) , 1-20
Open AccessSupervised Contrastive Models for Music Information Retrieval in Classical Persian Music
Musical instrument classification is a key task in music information retrieval, supporting applications such as automatic transcription and music recommendation. Research on this topic for traditional Persian music has been limited, largely due to the lack of complete and consistent datasets. This study introduces a reproducible baseline framework that combines supervised contrastive learning with stacked slice‑level aggregation—a late‑fusion approach that integrates predictions from short one‑second segments—to classify 15 traditional Persian instruments, including Ney, Setar, Tar, Santur, and Kamancheh. To support this work, we present the Persian Classical Instrument Dataset, a curated collection of publicly available solo‑instrument recordings encompassing 15 classical Persian instruments. Experiments across three settings—the five‑instrument subset, the full 15‑instrument dataset, and an existing baseline dataset—show that the proposed method achieves up to 99% accuracy on smaller subsets and 98% accuracy on the full dataset using 30‑second inputs. Furthermore, the same framework demonstrates strong generalization on the Dastgah detection task, outperforming previous methods and indicating that timbre‑based representations can transfer effectively to higher‑level modal recognition. Overall, this work provides both a publicly available dataset and a transparent, general‑purpose baseline to advance research on Persian and other non‑Western musical traditions.
- Top Cited PapersOpen Access
- 13 December 2025
- Published by Elsevier in Intelligent Systems with Applications
- Vol. 29
Vision transformers in precision agriculture: A comprehensive survey
- 13 December 2025
- Published by Elsevier in Intelligent Systems with Applications
- Vol. 29
Top Cited PapersOpen AccessVision transformers in precision agriculture: A comprehensive survey