arXiv subjects

Heshaam Faili

Publications and source records attributed to Heshaam Faili.

At least 19 recordsLinked to original sources

arXiv2026

Evaluating the Diagnostic Robustness of Vision-Language Models Under Visual and Textual Perturbations

Standard accuracy metrics for VLMs often mask significant reliability failures in sensitive domains. In this work, we utilize a histopathology-validated brain MRI dataset to systematically assess the diagnostic robustness of four VLM families under evidence-preserving perturbations. By reordering anatomical slices and swapping target label positions, we evaluate whether models maintain consistent predictions when clinical evidence remains invariant. Our results reveal significant vulnerabilities in presentation-order stability, with models exhibiting prediction flips in up to 48.9% of cases under simple sequence reversals. We further identify a textual selection bias, where label reordering triggers inconsistent diagnoses in up to 67.8% of cases despite identical visual inputs. Negative-control tests further reveal diagnostic overcommitment: models generate categorical diagnoses in up to 76.1% of cases after expert-annotated lesion slices are removed. These results demonstrate that high accuracy can overestimate clinical reliability, masking sensitivity to sequential presentation and textual framing that is not captured by aggregate accuracy. Our findings highlight the necessity of stability-based metrics for the deployment of VLMs in safety-critical clinical applications. Our evaluation data and code will be made public upon acceptance.

cs.CV↗

arXiv2026

MDP-GRPO: Stabilized Group Relative Policy Optimization for Multi-Constraint Instruction Following

Reinforcement learning with verifiable rewards is ideal for multi-constraint instruction following, yet standard group-relative policy optimization (GRPO) becomes unstable under discrete, low-dispersion rewards, where within-group reward distributions are frequently homogeneous. We identify and formalize three pathologies of z-score group normalization in this regime: low-variance amplification, mean-centering blindness, and zero-variance collapse. To address them, we propose MDP-GRPO, which stabilizes learning through (1) multi-temperature sampling to increase reward dispersion, (2) dual-anchor advantages to restore gradients in homogeneous groups and stop mean-centering blindness, (3) prospect-theoretic shaping to bound updates and penalize violations based on Kahneman and Tversky's theory, and (4) asymmetric KL regularization. Evaluated on FollowBench, IFEval, and a curated multi-constraint dataset, MDP-GRPO outperforms standard GRPO, improving strict constraint satisfaction by up to 5.0% on Llama-3.2-3B. Our method also enables stable convergence with small group sizes while preserving general capabilities on MMLU and ARC.

cs.LG↗

arXiv2026

PARSA-Bench: A Comprehensive Persian Audio-Language Model Benchmark

Persian poses unique audio understanding challenges through its classical poetry, traditional music, and pervasive code-switching - none captured by existing benchmarks. We introduce PARSA-Bench (Persian Audio Reasoning and Speech Assessment Benchmark), the first benchmark for evaluating large audio-language models on Persian language and culture, comprising 16 tasks and over 8,000 samples across speech understanding, paralinguistic analysis, and cultural audio understanding. Ten tasks are newly introduced, including poetry meter and style detection, traditional Persian music understanding, and code-switching detection. Text-only baselines consistently outperform audio counterparts, suggesting models may not leverage audio-specific information beyond what transcription alone provides. Culturally-grounded tasks expose a qualitatively distinct failure mode: all models perform near random chance on vazn detection regardless of scale, suggesting prosodic perception remains beyond the reach of current models. The dataset is publicly available at https://huggingface.co/datasets/MohammadJRanjbar/PARSA-Bench

cs.CL↗

arXiv2026

PersianPunc: A Large-Scale Dataset and BERT-Based Approach for Persian Punctuation Restoration

Punctuation restoration is essential for improving the readability and downstream utility of automatic speech recognition (ASR) outputs, yet remains underexplored for Persian despite its importance. We introduce PersianPunc, a large-scale, high-quality dataset of 17 million samples for Persian punctuation restoration, constructed through systematic aggregation and filtering of existing textual resources. We formulate punctuation restoration as a token-level sequence labeling task and fine-tune ParsBERT to achieve strong performance. Through comparative evaluation, we demonstrate that while large language models can perform punctuation restoration, they suffer from critical limitations: over-correction tendencies that introduce undesired edits beyond punctuation insertion (particularly problematic for speech-to-text pipelines) and substantially higher computational requirements. Our lightweight BERT-based approach achieves a macro-averaged F1 score of 91.33% on our test set while maintaining efficiency suitable for real-time applications. We make our dataset (https://huggingface.co/datasets/MohammadJRanjbar/persian-punctuation-restoration) and model (https://huggingface.co/MohammadJRanjbar/parsbert-persian-punctuation) publicly available to facilitate future research in Persian NLP and provide a scalable framework applicable to other morphologically rich, low-resource languages.

cs.CL↗

arXiv2025

ParsVoice: A Large-Scale Multi-Speaker Persian Speech Corpus for Text-to-Speech Synthesis

Persian remains substantially underrepresented in open speech-text resources, limiting progress in multi-speaker text-to-speech (TTS), speech-language modelling, and low-resource speech processing. We introduce ParsVoice, the largest publicly available Persian speech-text corpus tailored for training multi-speaker TTS systems, along with a scalable pipeline to construct high-quality speech-text data from long-form audiobook recordings. The pipeline combines a fine-tuned ParsBERT sentence-completion classifier, ASR-based boundary optimization, punctuation restoration, speaker identification, and a multi-dimensional quality assessment that covers both audio and Persian-specific text properties. The resulting release contains a 2,200-hour TTS-ready subset with 1.36 million aligned segments from 1,815 automatically identified speaker IDs, making it more than 25 times larger than the previously largest open Persian TTS dataset. To validate the corpus, we fine-tune XTTS, a zero-shot multilingual TTS model that operates directly on raw Persian text without phoneme representations, achieving a naturalness MOS of 3.6/5 and speaker similarity MOS of 4.0/5. The ParsVoice dataset is publicly available at: https://huggingface.co/datasets/MohammadJRanjbar/ParsVoice.

cs.SD↗

arXiv2025

SearchInstruct: Enhancing Domain Adaptation via Retrieval-Based Instruction Dataset Creation

Supervised Fine-Tuning (SFT) is essential for training large language models (LLMs), significantly enhancing critical capabilities such as instruction following and in-context learning. Nevertheless, creating suitable training datasets tailored for specific domains remains challenging due to unique domain constraints and data scarcity. In this paper, we propose SearchInstruct, an innovative method explicitly designed to construct high quality instruction datasets for SFT. Our approach begins with a limited set of domain specific, human generated questions, which are systematically expanded using a large language model. Subsequently, domain relevant resources are dynamically retrieved to generate accurate and contextually appropriate answers for each augmented question. Experimental evaluation demonstrates that SearchInstruct enhances both the diversity and quality of SFT datasets, leading to measurable improvements in LLM performance within specialized domains. Additionally, we show that beyond dataset generation, the proposed method can also effectively facilitate tasks such as model editing, enabling efficient updates to existing models. To facilitate reproducibility and community adoption, we provide full implementation details, the complete set of generated instruction response pairs, and the source code in a publicly accessible Git repository: [https://github.com/mostafaamiri/SearchInstruct](https://github.com/mostafaamiri/SearchInstruct)

cs.CL↗

arXiv2025

DeepQuestion: Systematic Generation of Real-World Challenges for Evaluating LLMs Performance

While Large Language Models (LLMs) achieve near-human performance on standard benchmarks, their capabilities often fail to generalize to complex, real-world problems. To bridge this gap, we introduce DeepQuestion, a scalable, automated framework that systematically elevates the cognitive complexity of existing datasets. Grounded in Bloom's taxonomy, DeepQuestion generates (1) scenario-based problems to test the application of knowledge in noisy, realistic contexts, and (2) instruction-based prompts that require models to create new questions from a given solution path, assessing synthesis and evaluation skills. Our extensive evaluation across ten leading open-source and proprietary models reveals a stark performance decline with accuracy dropping by up to 70% as tasks ascend the cognitive hierarchy. These findings underscore that current benchmarks overestimate true reasoning abilities and highlight the critical need for cognitively diverse evaluations to guide future LLM development.

cs.CL↗

arXiv2025

PersianMedQA: Evaluating Large Language Models on a Persian-English Bilingual Medical Question Answering Benchmark

Large Language Models (LLMs) have achieved remarkable performance on a wide range of Natural Language Processing (NLP) benchmarks, often surpassing human-level accuracy. However, their reliability in high-stakes domains such as medicine, particularly in low-resource languages, remains underexplored. In this work, we introduce PersianMedQA, a large-scale dataset of 20,785 expert-validated multiple-choice Persian medical questions from 14 years of Iranian national medical exams, spanning 23 medical specialties and designed to evaluate LLMs in both Persian and English. We benchmark 41 state-of-the-art models, including general-purpose, Persian, and medical LLMs, in zero-shot and chain-of-thought (CoT) settings. Our results show that closed-weight general models (e.g., GPT-4.1) consistently outperform all other categories, achieving 83.09% accuracy in Persian and 80.7% in English, while Persian LLMs such as Dorna underperform significantly (e.g., 34.9% in Persian), often struggling with both instruction-following and domain reasoning. We also analyze the impact of translation, showing that while English performance is generally higher, 3-10% of questions can only be answered correctly in Persian due to cultural and clinical contextual cues that are lost in translation. Finally, we demonstrate that model size alone is insufficient for robust performance without strong domain or language adaptation. PersianMedQA provides a foundation for evaluating bilingual and culturally grounded medical reasoning in LLMs. The dataset, along with a bilingual medical dictionary, is available: https://huggingface.co/datasets/MohammadJRanjbar/PersianMedQA .

cs.CL↗

arXiv2025

SchemaGraphSQL: Efficient Schema Linking with Pathfinding Graph Algorithms for Text-to-SQL on Large-Scale Databases

Text-to-SQL systems translate natural language questions into executable SQL queries, and recent progress with large language models (LLMs) has driven substantial improvements in this task. Schema linking remains a critical component in Text-to-SQL systems, reducing prompt size for models with narrow context windows and sharpening model focus even when the entire schema fits. We present a zero-shot, training-free schema linking approach that first constructs a schema graph based on foreign key relations, then uses a single prompt to Gemini 2.5 Flash to extract source and destination tables from the user query, followed by applying classical path-finding algorithms and post-processing to identify the optimal sequence of tables and columns that should be joined, enabling the LLM to generate more accurate SQL queries. Despite being simple, cost-effective, and highly scalable, our method achieves state-of-the-art results on the BIRD benchmark, outperforming previous specialized, fine-tuned, and complex multi-step LLM-based approaches. We conduct detailed ablation studies to examine the precision-recall trade-off in our framework. Additionally, we evaluate the execution accuracy of our schema filtering method compared to other approaches across various model sizes.

cs.CL↗

arXiv2025

Matina: A Large-Scale 73B Token Persian Text Corpus

Text corpora are essential for training models used in tasks like summarization, translation, and large language models (LLMs). While various efforts have been made to collect monolingual and multilingual datasets in many languages, Persian has often been underrepresented due to limited resources for data collection and preprocessing. Existing Persian datasets are typically small and lack content diversity, consisting mainly of weblogs and news articles. This shortage of high-quality, varied data has slowed the development of NLP models and open-source LLMs for Persian. Since model performance depends heavily on the quality of training data, we address this gap by introducing the Matina corpus, a new Persian dataset of 72.9B tokens, carefully preprocessed and deduplicated to ensure high data quality. We further assess its effectiveness by training and evaluating transformer-based models on key NLP tasks. Both the dataset and preprocessing codes are publicly available, enabling researchers to build on and improve this resource for future Persian NLP advancements.

cs.CL↗

arXiv2024

PerSHOP -- A Persian dataset for shopping dialogue systems modeling

Nowadays, dialogue systems are used in many fields of industry and research. There are successful instances of these systems, such as Apple Siri, Google Assistant, and IBM Watson. Task-oriented dialogue system is a category of these, that are used in specific tasks. They can perform tasks such as booking plane tickets or making restaurant reservations. Shopping is one of the most popular areas on these systems. The bot replaces the human salesperson and interacts with the customers by speaking. To train the models behind the scenes of these systems, annotated data is needed. In this paper, we developed a dataset of dialogues in the Persian language through crowd-sourcing. We annotated these dialogues to train a model. This dataset contains nearly 22k utterances in 15 different domains and 1061 dialogues. This is the largest Persian dataset in this field, which is provided freely so that future researchers can use it. Also, we proposed some baseline models for natural language understanding (NLU) tasks. These models perform two tasks for NLU: intent classification and entity extraction. The F-1 score metric obtained for intent classification is around 91% and for entity extraction is around 93%, which can be a baseline for future research.

cs.CL↗

arXiv2023

Persian Typographical Error Type Detection Using Deep Neural Networks on Algorithmically-Generated Misspellings

Spelling correction is a remarkable challenge in the field of natural language processing. The objective of spelling correction tasks is to recognize and rectify spelling errors automatically. The development of applications that can effectually diagnose and correct Persian spelling and grammatical errors has become more important in order to improve the quality of Persian text. The Typographical Error Type Detection in Persian is a relatively understudied area. Therefore, this paper presents a compelling approach for detecting typographical errors in Persian texts. Our work includes the presentation of a publicly available dataset called FarsTypo, which comprises 3.4 million words arranged in chronological order and tagged with their corresponding part-of-speech. These words cover a wide range of topics and linguistic styles. We develop an algorithm designed to apply Persian-specific errors to a scalable portion of these words, resulting in a parallel dataset of correct and incorrect words. By leveraging FarsTypo, we establish a strong foundation and conduct a thorough comparison of various methodologies employing different architectures. Additionally, we introduce a groundbreaking Deep Sequential Neural Network that utilizes both word and character embeddings, along with bidirectional LSTM layers, for token classification aimed at detecting typographical errors across 51 distinct classes. Our approach is contrasted with highly advanced industrial systems that, unlike this study, have been developed using a diverse range of resources. The outcomes of our final method proved to be highly competitive, achieving an accuracy of 97.62%, precision of 98.83%, recall of 98.61%, and surpassing others in terms of speed.

cs.CL↗

arXiv2022

Mismatching-Aware Unsupervised Translation Quality Estimation For Low-Resource Languages

Translation Quality Estimation (QE) is the task of predicting the quality of machine translation (MT) output without any reference. This task has gained increasing attention as an important component in the practical applications of MT. In this paper, we first propose XLMRScore, which is a cross-lingual counterpart of BERTScore computed via the XLM-RoBERTa (XLMR) model. This metric can be used as a simple unsupervised QE method, nevertheless facing two issues: firstly, the untranslated tokens leading to unexpectedly high translation scores, and secondly, the issue of mismatching errors between source and hypothesis tokens when applying the greedy matching in XLMRScore. To mitigate these issues, we suggest replacing untranslated words with the unknown token and the cross-lingual alignment of the pre-trained model to represent aligned words closer to each other, respectively. We evaluate the proposed method on four low-resource language pairs of the WMT21 QE shared task, as well as a new English$\rightarrow$Persian (En-Fa) test dataset introduced in this paper. Experiments show that our method could get comparable results with the supervised baseline for two zero-shot scenarios, i.e., with less than 0.01 difference in Pearson correlation, while outperforming unsupervised rivals in all the low-resource language pairs for above 8%, on average.

cs.CL↗

arXiv2022

Pruned Graph Neural Network for Short Story Ordering

Text coherence is a fundamental problem in natural language generation and understanding. Organizing sentences into an order that maximizes coherence is known as sentence ordering. This paper is proposing a new approach based on the graph neural network approach to encode a set of sentences and learn orderings of short stories. We propose a new method for constructing sentence-entity graphs of short stories to create the edges between sentences and reduce noise in our graph by replacing the pronouns with their referring entities. We improve the sentence ordering by introducing an aggregation method based on majority voting of state-of-the-art methods and our proposed one. Our approach employs a BERT-based model to learn semantic representations of the sentences. The results demonstrate that the proposed method significantly outperforms existing baselines on a corpus of short stories with a new state-of-the-art performance in terms of Perfect Match Ratio (PMR) and Kendall's Tau (Tau) metrics. More precisely, our method increases PMR and Tau criteria by more than 5% and 4.3%, respectively. These outcomes highlight the benefit of forming the edges between sentences based on their cosine similarity. We also observe that replacing pronouns with their referring entities effectively encodes sentences in sentence-entity graphs.

cs.CL↗

arXiv2021

A New Sentence Ordering Method Using BERT Pretrained Model

Building systems with capability of natural language understanding (NLU) has been one of the oldest areas of AI. An essential component of NLU is to detect logical succession of events contained in a text. The task of sentence ordering is proposed to learn succession of events with applications in AI tasks. The performance of previous works employing statistical methods is poor, while the neural networks-based approaches are in serious need of large corpora for model learning. In this paper, we propose a method for sentence ordering which does not need a training phase and consequently a large corpus for learning. To this end, we generate sentence embedding using BERT pre-trained model and measure sentence similarity using cosine similarity score. We suggest this score as an indicator of sequential events' level of coherence. We finally sort the sentences through brute-force search to maximize overall similarities of the sequenced sentences. Our proposed method outperformed other baselines on ROCStories, a corpus of 5-sentence human-made stories. The method is specifically more efficient than neural network-based methods when no huge corpus is available. Among other advantages of this method are its interpretability and needlessness to linguistic knowledge.

cs.CL↗

arXiv2021

NLP-IIS@UT at SemEval-2021 Task 4: Machine Reading Comprehension using the Long Document Transformer

This paper presents a technical report of our submission to the 4th task of SemEval-2021, titled: Reading Comprehension of Abstract Meaning. In this task, we want to predict the correct answer based on a question given a context. Usually, contexts are very lengthy and require a large receptive field from the model. Thus, common contextualized language models like BERT miss fine representation and performance due to the limited capacity of the input tokens. To tackle this problem, we used the Longformer model to better process the sequences. Furthermore, we utilized the method proposed in the Longformer benchmark on Wikihop dataset which improved the accuracy on our task data from 23.01% and 22.95% achieved by the baselines for subtask 1 and 2, respectively, to 70.30% and 64.38%.

cs.CL↗

arXiv2020

Cross-Lingual Adaptation Using Universal Dependencies

We describe a cross-lingual adaptation method based on syntactic parse trees obtained from the Universal Dependencies (UD), which are consistent across languages, to develop classifiers in low-resource languages. The idea of UD parsing is to capture similarities as well as idiosyncrasies among typologically different languages. In this paper, we show that models trained using UD parse trees for complex NLP tasks can characterize very different languages. We study two tasks of paraphrase identification and semantic relation extraction as case studies. Based on UD parse trees, we develop several models using tree kernels and show that these models trained on the English dataset can correctly classify data of other languages e.g. French, Farsi, and Arabic. The proposed approach opens up avenues for exploiting UD parsing in solving similar cross-lingual tasks, which is very useful for languages that no labeled data is available for them.

cs.CL↗

arXiv2020

NSURL-2019 Task 7: Named Entity Recognition (NER) in Farsi

NSURL-2019 Task 7 focuses on Named Entity Recognition (NER) in Farsi. This task was chosen to compare different approaches to find phrases that specify Named Entities in Farsi texts, and to establish a standard testbed for future researches on this task in Farsi. This paper describes the process of making training and test data, a list of participating teams (6 teams), and evaluation results of their systems. The best system obtained 85.4% of F1 score based on phrase-level evaluation on seven classes of NEs including person, organization, location, date, time, money and percent.

cs.CL↗