Quantum-Enhanced Handwritten Bangla Character Recognition: A Hybrid Quantum Classical Neural Network Approach

Quantum-Enhanced Handwritten Bangla Character Recognition: A Hybrid Quantum Classical Neural Network Approach | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Quantum-Enhanced Handwritten Bangla Character Recognition: A Hybrid Quantum Classical Neural Network Approach Md Abdur Rahman, Raja Tariqul Hasan Tusher, Md Rageb Shakil Hridoy, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8187167/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Handwritten Bangla character recognition poses a significant computer vision challenge for over 250 million native speakers. The script's morphological complexity, which includes a large set of basic characters, numerals, modifiers, and intricate compound conjuncts, demands computationally robust models. While classical Convolutional Neural Networks (CNNs) have shown high accuracy, they are often demanding and struggle with subtle character variations. This paper fills a critical research gap by designing and evaluating, to our knowledge, the first Hybrid Quantum-Classical Convolutional Neural Network (HQCNN) for this task. We propose a novel architecture that integrates a quatum convolutional layer, implemented using Random Quantum Circuits (RQCs) simulated with PennyLane, as a high-dimensional feature extractor. To isolate the impact of the quantum layer, we conducted a rigorous comparative analysis of our HQCNN against a structurally identical classical CNN baseline across seven distinct experiments on four public datasets: NumtaDB, CMATERdb 3.1.2, Ekush, and BanglaLekha-Isolated. The HQCNN consistently outperformed the classical baseline in all seven tasks, achieving a peak accuracy of 99.45% on the Ekush numerical dataset. Notably, the most significant "quantum advantage" was observed in the classification of structurally complex compound characters (E6), where the HQCNN achieved 97.16% accuracy versus the baseline's 95.52% (a 1.64% improvement). Furthermore, the HQCNN demonstrated superior learning efficiency and stability. The quantum-derived features allowed the classical backbone to converge 27–43% faster (e.g., 162.78 vs. 223.37 minutes of classical backbone training time on the 84-class mixed dataset) and with a more stable validation loss. These results provide strong evidence that the quantum convolutional layer captures more expressive feature representations, demonstrating that the static RQC layer that the static RQC layer, by operating in a high-dimensional Hilbert space, provides a more expressive feature representation than a classical kernel, making the classification task easier for the classical backbone. While classical simulation of the quantum circuits adds significant computational overhead, our findings on faster convergence strongly suggest that an implementation on native quantum hardware would offer a substantial wall-clock speedup, presenting a viable and highly efficient paradigm for advancing OCR systems for Bangla and other complex Indic scripts. Hybrid Quantum-Classical Convolutional Neural Network Bangla Handwritten Character Recognition Quantum Machine Learning Convolutional Neural Networks Random Quantum Circuits PennyLane Quantum Convolution Feature Extraction Multilingual OCR Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8187167","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":550423728,"identity":"6591464f-9298-4751-9c71-b2d72ea3f8d2","order_by":0,"name":"Md Abdur Rahman","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBElEQVRIiWNgGAWjYFACHhAhUc/AcPyA4YcKZgiXgeEAYwN+LRYJDIxnEoolzhCvpSKBgfmAwQfeNmYGglrM288e3fBzh0QeA9uBxA2S86xlDA4wH7zNw3BHFpcWmTN5aTd7z0gUM/AcPGxQuC2dx+AAW7I1D8MzY1xaJBhyzG7wtkkw7r9xIM1ActthoBYeM2kehsOJOLXwvzG7+ReopUH+gfkP3jkgLfzf8GuRyDG7DbQFqOCAgQFvA9gWNgJa3pjdlm2TMGZgOJNgLHEsnUfyMJux5RwDPH7hzzG7+batTg4SlTXW9nzHmx/eeFOBO8SwAHDUGBCvfhSMglEwCkYBJgAAN75aiJj/NyEAAAAASUVORK5CYII=","orcid":"","institution":"Daffodil International University","correspondingAuthor":true,"prefix":"","firstName":"Md","middleName":"Abdur","lastName":"Rahman","suffix":""},{"id":550423729,"identity":"df7cde82-cfb0-40a6-b968-b0a3de45cd1d","order_by":1,"name":"Raja Tariqul Hasan Tusher","email":"","orcid":"","institution":"Daffodil International University","correspondingAuthor":false,"prefix":"","firstName":"Raja","middleName":"Tariqul Hasan","lastName":"Tusher","suffix":""},{"id":550423730,"identity":"5b1fa00c-2f62-4da7-a946-920ebde8db18","order_by":2,"name":"Md Rageb Shakil Hridoy","email":"","orcid":"","institution":"Daffodil International University","correspondingAuthor":false,"prefix":"","firstName":"Md","middleName":"Rageb Shakil","lastName":"Hridoy","suffix":""},{"id":550423731,"identity":"aefdd09b-3a71-422d-8e8b-90020cc852ab","order_by":3,"name":"Md. Sadekur Rahman","email":"","orcid":"","institution":"Daffodil International University","correspondingAuthor":false,"prefix":"","firstName":"Md.","middleName":"Sadekur","lastName":"Rahman","suffix":""}],"badges":[],"createdAt":"2025-11-23 18:08:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8187167/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8187167/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":101881219,"identity":"97beeb33-15c1-4477-abb4-3fe84139a546","added_by":"auto","created_at":"2026-02-04 15:10:51","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2695383,"visible":true,"origin":"","legend":"","description":"","filename":"splnproc1703.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8187167/v1_covered_95433856-91cc-4a3d-90be-1da1716accdb.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Quantum-Enhanced Handwritten Bangla Character Recognition: A Hybrid Quantum Classical Neural Network Approach","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Hybrid Quantum-Classical Convolutional Neural Network, Bangla Handwritten Character Recognition, Quantum Machine Learning, Convolutional Neural Networks, Random Quantum Circuits, PennyLane, Quantum Convolution, Feature Extraction, Multilingual OCR","lastPublishedDoi":"10.21203/rs.3.rs-8187167/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8187167/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eHandwritten Bangla character recognition poses a significant computer vision challenge for over 250\u0026nbsp;million native speakers. The script's morphological complexity, which includes a large set of basic characters, numerals, modifiers, and intricate compound conjuncts, demands computationally robust models. While classical Convolutional Neural Networks (CNNs) have shown high accuracy, they are often demanding and struggle with subtle character variations. This paper fills a critical research gap by designing and evaluating, to our knowledge, the first Hybrid Quantum-Classical Convolutional Neural Network (HQCNN) for this task. We propose a novel architecture that integrates a quatum convolutional layer, implemented using Random Quantum Circuits (RQCs) simulated with PennyLane, as a high-dimensional feature extractor. To isolate the impact of the quantum layer, we conducted a rigorous comparative analysis of our HQCNN against a structurally identical classical CNN baseline across seven distinct experiments on four public datasets: NumtaDB, CMATERdb 3.1.2, Ekush, and BanglaLekha-Isolated. The HQCNN consistently outperformed the classical baseline in all seven tasks, achieving a peak accuracy of 99.45% on the Ekush numerical dataset. Notably, the most significant \"quantum advantage\" was observed in the classification of structurally complex compound characters (E6), where the HQCNN achieved 97.16% accuracy versus the baseline's 95.52% (a 1.64% improvement). Furthermore, the HQCNN demonstrated superior learning efficiency and stability. The quantum-derived features allowed the classical backbone to converge 27\u0026ndash;43% faster (e.g., 162.78 vs. 223.37 minutes of classical backbone training time on the 84-class mixed dataset) and with a more stable validation loss. These results provide strong evidence that the quantum convolutional layer captures more expressive feature representations, demonstrating that the static RQC layer that the static RQC layer, by operating in a high-dimensional Hilbert space, provides a more expressive feature representation than a classical kernel, making the classification task easier for the classical backbone. While classical simulation of the quantum circuits adds significant computational overhead, our findings on faster convergence strongly suggest that an implementation on native quantum hardware would offer a substantial wall-clock speedup, presenting a viable and highly efficient paradigm for advancing OCR systems for Bangla and other complex Indic scripts.\u003c/p\u003e","manuscriptTitle":"Quantum-Enhanced Handwritten Bangla Character Recognition: A Hybrid Quantum Classical Neural Network Approach","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-03 11:20:38","doi":"10.21203/rs.3.rs-8187167/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"38876366-f0f0-486a-a025-7f6abbef5c05","owner":[],"postedDate":"February 3rd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-02-03T11:20:39+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-03 11:20:38","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8187167","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8187167","identity":"rs-8187167","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}