Mova0.5 pre-release⌕
☰
home / docs / papers / Ancient Hebrew / Deep Aramaic: Towards a Synthetic Data Paradigm Enabling Machine Learning in Epigraphy

Deep Aramaic: Towards a Synthetic Data Paradigm Enabling Machine Learning in Epigraphy

What it is about

Labelled images of Old Aramaic letters are too few to train a letter classifier, so the authors render synthetic ones: 22 letter classes drawn in the styles of inscriptions, set on simulated stone, ceramic or papyrus textures, lit from varied angles and damaged by erosion and breakage, 250 000 training and 25 000 validation images. A ResNet trained only on these reaches about 95% accuracy on real letters of the 8th-century BCE Hadad statue (KAI 214) and similar results on other materials; accuracy grows with the size and variety of the synthetic set (about 92% at 150 000 images, 98% at 500 000). The work is about images, not language.

Ideas

For Mova

Mova does not read images, so the classifier itself is out of scope. I1 is a test harness for macula dict lookup and a future unvocalised reader; I2 is a short note in 1. Script and pointing (letter shapes, scribal confusions). The yetzirah crate's letter tables are not affected.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. arXiv 2310.07310

@misc{aioanei2023deep,
  title = {{Deep Aramaic: Towards a Synthetic Data Paradigm Enabling Machine Learning in Epigraphy}},
  author = {Aioanei, Andrei C. and Hunziker-Rodewald, Regine and Klein, Konstantin and Michels, Dominik L.},
  year = {2023},
  eprint = {2310.07310},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2310.07310}
}

References

7 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G. S., Davis, A., Dean, J., Devin, M., Ghemawat, S., Goodfellow, I., Harp, A., Irving, G., Isard, M., Jia, Y., Jozefowicz, R., Kaiser, L., Kudlur, M., Levenberg, J., ... , and Zheng, X. (2015). TensorFlow: Large-scale machine learning on heterogeneous systems. Retrieved from http://tensorflow.org. + [ADDG 21] Anichini, F., Dershowitz, N., Dubbini, N., Gattiglia, G., Itkin, B. and Wolf, L. (2021). The automatic recognition of ceramics from only one photo: The ArchAIDE app. Journal of Archaeological Science: Reports 36, 102788. 10.1016/j.jasrep.2020.102788. [AJGV21] Abbas, A., Jain, S., Gour, M. and Vankudothu, S. (2021). Tomato plant disease detection using transfer learning with C-GAN synthetic images. Computers and Electronics in Agriculture 187. 10.1016/j.compag.2021.106279. [Arc18] Arcidiacono, C. S. (2018). An empirical study on synthetic image generation techniques for object detectors, Phd Dissertation, KTH, School of Electrical Engineering and Computer Science (EECS). [ASM21] Arnia, F., Saddami, K. and Munadi, K. (2021). DCNet: Noise-Robust Convolutional Neural Networks for Degradation Classification on Ancient Documents. Journal of Imaging, 7(7), 114. https://doi.org/10.3390/jimaging7070114 [ASP19] Assael, Y., Sommerschield, T. and Prag, J. (2019). Restoring ancient text using deep learning: a case study on Greek epigraphy. Conference on Empirical Methods in Natural Language Processing. + [ASSB 22] Assael, Y., Sommerschield, T., Shillingford, B., Bordbar, M., Pavlopoulos, J., Chatzipanagiotou, M., Androutsopoulos, I., Prag, J. and de Freitas, N. (2022). Restoring and attributing ancient texts using deep neural networks. Nature 603(7900), 280–83. https://doi.org/10.1038/s41586-022-04448-z. + [BCCF 22] Barucci, A., Canfailla, C., Cucci, C., Forasassi, M., Franci, M., Guarducci, G., Guidi, T., Loschiavo, M., Picollo, M., Pini, R., Python, L., Valentini, S. and Argenti, F. (2022). Ancient Egyptian Hieroglyphs Segmentation and Classification with Convolutional Neural Networks. In R. Furferi, L. Governi, Y. Volpe, K. Seymour, A. Pelagotti and F. Gherardini (eds), The Future of Heritage Science and Technologies: ICT and Digital Heritage. Florence Heri-Tech 2022. Communications in Computer and Information Science, vol. 1645. Springer, Cham. https://doi.org/10.1007/978-3-031-20302-2_10. + [BCFL 21] Barucci, A., Cucci, C., Franci, M., Loschiavo, M. and Argenti, F. (2021). A Deep Learning Approach to Ancient Egyptian Hieroglyphs Classification. IEEE Access 9, 123438–47. doi: 10.1109/ACCESS.2021.3110082. [BGK22] Bermeitinger, B., Gulden, S. A. and Konrad, T (2022). How to Compute a Shape: Optical Character Recognition for Hieratic. 10.25358/openscience-6757. [Bir18] Birnbaum, S. (2018). The Old Aramaic Inscription of Tel Dan. The Ostracon 28(2), 3–14. [BM22] Bogacz, B. and Mara, H. (2022). Digital Assyriology—Advances in Visual Cuneiform Analysis. Journal on Computing and Cultural Heritage (JOCCH) 15, 1–22. [BP09] Bordreuil, P. and Pardee, D. (2009). A Manual of Ugaritic. Winona Lake, IN: Eisenbrauns. [BPH22] Burdorf, S., Plum, K., and Hasenklever, D. (2022). Reducing the Amount of Real World Data for Object Detector Training with Synthetic Data. ArXiv, abs/2202.00632. [CBHC19] Choi, G., Bagheri, M. A., Handali, J. and Choi, H. (2019). Generating synthetic handwritten data using generative adversarial networks. arXiv preprint arXiv:1902.01077. [CH23] Chollet, F. and Howard, A. G. (2023). Image Processing Neural Networks With Separable Convolutional Layers. [CO15] Chollet, F. et al. (2015). Keras. GitHub. Retrieved from https://github.com/fchollet/keras. [COG18] Can, G., Odobez, J.-M. and Gatica-Perez, D. (2018). How to Tell Ancient Signs Apart? Recognizing and Visualizing Maya Glyphs with CNNs. Journal on Computing and Cultural Heritage 11, 1–25. 10.1145/3230670. [CTVF22] Corazza M., Tamburini F., Valério M. & Ferrara, S. (2022). Unsupervised deep learning supports reclassification of Bronze age cypriot writing system. PLOS ONE 17(7), e0269544. https://doi.org/10.1371/journal.pone.0269544. + [DDSL 09] Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K. and Fei-Fei, L. (2009). ImageNet: A Large-Scale Hierarchical Image Database. IEEE Conference on Computer Vision and Pattern Recognition, Miami, 20-25 June 2009, 248–55. https://doi.org/10.1109/CVPR.2009.5206848. ↗
  2. Duque-Domingo, J., Herrera, P. J., Valero, E. and Cerrada, C. (2017). Deciphering Egyptian Hieroglyphs: Towards a New Strategy for Navigation in Museums. Sensors 17, 589. doi:10.3390/s17030589. [DR69] Donner, H. and Röllig, W. (1966-1969). Kanaanäische und Aramäische Inschriften. Wiesbaden: Harrassowitz. [EL21] Eversberg, L. and Lambrecht, J. (2021). Generating Images with Physics-Based Rendering for an Industrial Object Detection Task: Realism versus Domain Randomization. Sensors 21, 7901. 10.3390/s21237901. [Ferr20] Ferrara, S. (2020). Icon, index, symbol: Language notation in the Cretan Hieroglyphic script. In J. Bennet (ed), Representations: Material and immaterial modes of communication in the Bronze Age Aegean, Oxbow, Oxford, 211–40. [Fit95] Fitzmyer, J. A. (1995). The Aramaic Inscriptions of Sefire. Revised Edition. Rome: Pontifical Biblical Institute. [FLAG20] Fetaya, E., Lifshitz, Y., Aaron, E. and Gordin, S. (2020). Restoration of fragmentary Babylonian texts using recurrent neural networks. Proceedings of the National Academy of Sciences of the United States of America 117(37), 22743–51. https://doi.org/10.1073/pnas.2003794117. + [FSSF 15] Faigenbaum-Golovin, S., Shaus, A., Sober, B., Finkelstein, I., Levin, D., Piasetzky, E and Turkel, E. (2015). Computerized Paleographic Investigation of Hebrew Iron Age Ostraca. Radiocarbon 57, 317–25. 10.2458/azu_rc.57.18565. [FTNZ+21] Fedorova, S., Tono, A., Nigam, M. S., Zhang, J., Ahmadnia, A., Bolognesi, C. and Michels, D. (2021). Synthetic 3D Data Generation Pipeline for Geometric Deep Learning in Architecture. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XLIII-B2-2021, 337–44. doi:10.5194/isprs-archives-xliii-b2-2021-337-2021. [FvG13] Franken, M. and van Gemert, J. (2013). Automatic Egyptian hieroglyph recognition by retrieving images as texts, Proceedings of the 21st ACM international conference on Multimedia, 765–68. [GGA21] Gualandi, M. L., Gattiglia, G. and Anichini, F. (2021). An Open System for Collection and Automatic Recognition of Pottery through Neural Network Algorithms. Heritage 4(1), 140–59. https://doi.org/10.3390/heritage4010008. [Gib71] Gibson, J.C.L. (1971). Textbook of Syrian Semitic Inscriptions: Aramaic Inscriptions Including Inscriptions in the Dialect of Zenjirli. Clarendon Press: Oxford. [GJ14] Graves, A. and Jaitly, N. (2014). Towards end-to-end speech recognition with recurrent neural networks. International conference on machine learning. PMLR. + [GPFC 23] Guidi, T., Python, L., Forasassi, M., Cucci, C., Franci, M., Argenti, F. and Barucci, A. (2023). Egyptian Hieroglyphs Segmentation with Convolutional Neural Networks. Algorithms 16(2), 79. 10.3390/a16020079. [GVZ16] Gupta, A., Vedaldi, A. and Zisserman, A. (2016). Synthetic data for text localisation in natural images. Proceedings of the IEEE conference on computer vision and pattern recognition. ↗ our card
  3. Gzella (2015). A Cultural History of Aramaic: From the Beginnings to the Advent of Islam. Leiden, Netherlands: Brill. + [HYDH 22] Huang, H., Yang, D., Dai, G., Han, Z., Wang, Y., Lam, K., Yang, F., Huang, S., Liu, Y., and He, M. (2022). AGTGAN: Unpaired Image Translation for Photographic Ancient Character Generation. Proceedings of the 30th ACM International Conference on Multimedia. [HZRS16] He, K., Zhang, X., Ren, S. and Sun, J. (2016). Identity Mappings in Deep Residual Networks. ECCV, 630–45. [IWD21] Itkin, B., Wolf, L. and Dershowitz, N. (2021). Computational Visual Ceramicology: Matching Image Outlines to Catalog Sketches. Proceedings of the AAAI Conference on Artificial Intelligence, 35(17), 14822–30. https://doi.org/10.1609/aaai.v35i17.17740. + [JBMN 17] Johnson-Roberson, M., Barto, C., Mehta, R., Nittur, S. Sridhar, Rosaen, K. and Vasudevan, R. (2017). Driving in the matrix: Can virtual worlds replace human-generated annotations for real world tasks?. arXiv preprint arXiv:1709.07791. [JSVZ14] Jaderberg, M, Simonyan, K., Vedaldi, A. and Zisserman, A. (2014). Synthetic data and artificial neural networks for natural scene text recognition. arXiv preprint arXiv:1406.2227. [KB14] Kingma, D. P. and Ba, J. (2014). Adam: A method for stochastic optimization. arXiv 2014, arXiv:1412.6980. + [KWPP 23] Klein, J., Waller, R. E., Pirk, S., Pałubicki, W., Tester, M. and Michels, D. (2023). Synthetic Data at Scale: A Paradigm to Efficiently Leverage Machine Learning in Agriculture. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4314564. [LNN18] Ly, N., Nguyen, C. and Nakagawa, M. (2018). Training an End-to-End Model for Offline Handwritten Japanese Text Recognition by Generated Synthetic Patterns. 16th International Conference on Frontiers in Handwriting Recognition (ICFHR), Niagara Falls, NY, USA, 74–9. doi: 10.1109/ICFHR-2018.2018.00022. [LS13] Lemaire, A. and Sass, B. (2013). The Mortuary Stele with Sam'alian Inscription from Ördekburnu near Zincirli. Bulletin of the American Schools of Oriental Research, No. 369, pp. 57–136. + [LSYH 21] Lazar, K., Saret, B., Yehudai, A., Horowitz, W., Wasserman, N. A. and Stanovsky, G. (2021). Filling the Gaps in Ancient Akkadian Texts: A Masked Language Modelling Approach. Conference on Empirical Methods in Natural Language Processing. https://doi.org/10.48550/arXiv.2109.04513. [LZJJ19] Li, J., Zhou, H., Jayas, D. and Jia, Q. (2019). Construction of a Dataset of Stored-grain Insects Images for Intelligent Monitoring. Applied Engineering in Agriculture 35, 647–55. 10.13031/aea.13122. + [MHSM 19] Makowski, M., Hädrich, T., Scheffczyk, J., Michels, D., Pirk, S. and Palubicki, W. (2019). Synthetic silviculture: multi-scale modeling of plant ecosystems. ACM Transactions on Graphics 38 (4), 1–14. 10.1145/3306346.3323039. ↗ our card
  4. Muraoka, T. and Porten, B. (1998). A Grammar of Egyptian Aramaic. Leiden, Netherlands: Brill. [Nav82] Naveh, J. (1982). The Development of the Aramaic Script. Proceedings of the Israel Academy of Sciences and Humanities 5(1), 32-57. [Nav87] Naveh, J. (1987). Early history of the alphabet: an introduction to West Semitic epigraphy and palaeography. Jerusalem: Magnes Press. [Nav986] Naveh, J. (1986). Some Considerations on the Priest Panammuwa's Inscription from Sam'al. Israel Exploration Journal 36(1/2), 27-32. [Nik21] Nikolenko, S. I. (2021). Synthetic Data for Deep Learning, Springer, Cham, Switzerland. + [NKKH 19] Nowruzi, F. E., Kapoor, P., Kolhatkar, D., Hassanat, F. A., Laganière, R. and Rebut, J. (2019). How much real data do we actually need: Analyzing object detection performance using synthetic and real data. ArXiv, abs/1907.07061. [Nöl01] Nöldeke, T (2001). Compendious Syriac Grammar. Winona Lake, IN: Eisenbrauns. [NvHMT + 21] Núñez Jareño, S. J., van Helden, D. P., Mirkes, E. M., Tyukin, I. Y. and Allison, P. M. (2021). Learning from Scarce Information: Using Synthetic Data to Classify Roman Fine Ware Pottery. Entropy 23(9), 1140. [PKO23] Papavassileiou, K., Kosmopoulos, D. I. and Owens, G. (2023). A generative model for the Mycenaean Linear B script and its application in infilling text from ancient tablets. ACM Journal on Computing and Cultural Heritage. 10.1145/3593431. [PMGB+23] Pellis, E., Masiero, A., Grussenmeyer, P., Betti, M. and Tucci, G. (2023). Synthetic Data Generation and Testing for the Semantic Segmentation of Heritage Buildings. Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVIII-M-2-2023, 1189–96. 10.5194/isprs-archives-XLVIII-M-2-2023-1189-2023. [RMPF+21] Ragusa, F., Mauro, D. D., Palermo, A., Furnari, A. and Farinella, G. M. (2021). Semantic Object Segmentation in Cultural Sites using Real and Synthetic Data. 2020 25th International Conference on Pattern Recognition (ICPR), 1964–71. [Ros15] Rosmorduc, S. (2015). Computational Linguistics in Egyptology. UCLA Encyclopedia of Egyptology. Hal-02982113. [Ros20] Rosmorduc, S. (2020). Automated Transliteration of Late Egyptian Using Neural Networks: An Experiment in ‘Deep Learning’, Lingua Aegyptia - Journal of Egyptian Language Studies 28, 233–57. 10.37011/lingaeg.28.07. [RRRK22] Rizk, R., Rizk, D., Rizk, F. and Kumar, A. (2021). A Hybrid Capsule Network-based Deep Learning Framework for Deciphering Ancient Scripts with Scarce Annotations: A Case Study on Phoenician Epigraphy. 2021 IEEE International Midwest Symposium on Circuits and Systems (MWSCAS), 617–20. + [SAPS 23] Sommerschield, T., Assael, Y., Pavlopoulos, J., Stefanak, V., Senior, A., Dyer, C., Bodel, J., Prag, J., Androutsopoulos, I. & Freitas, N. (2023). Machine Learning for Ancient Languages: A Survey. Computational Linguistics, 1–44. 10.1162/coli_a_00481 ↗
  5. Shaus, A., Gerber, Y., Faigenbaum-Golovin, S., Sober, B., Piasetzky, E. and Finkelstein, I. (2020). Forensic document examination and algorithmic handwriting analysis of Judahite biblical period inscriptions reveal significant literacy level. PLoS ONE 15. [Sha17] Shaus, A. (2017). Computer Vision and Machine Learning Methods for Analyzing First Temple Period Inscriptions, PhD Dissertation, Tel Aviv University. + [SLJS 15] Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V. and Rabinovich, A. (2015). Going deeper with convolutions. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, MA, USA, 1–9, doi: 10.1109/CVPR.2015.7298594. [Sok19] Sokoloff, M. (2019). The Jewish Babylonian Aramaic Bowls. Volume 1. Jerusalem: The Hebrew University Magnes Press. [SS11] Sader, H. and Schroeder O. (2011). Aramaic. In S. R. Steadman and G. McMahon (eds.), The Oxford Handbook of Ancient Anatolia. Oxford: Oxford University Press. [Sta20] Stadel, C. (2020). Old and Imperial Aramaic. In R. Hasselbach‐Andee (ed), A Companion to Ancient Near Eastern Languages, Wiley, Blackwell, NJ, 319–36. [Tei15] Teixidor, J. (2015). The Aramaic Language in the Achaemenid Period. Leiden, Netherlands: Brill. + [TFRS 17] Tobin, J., Fong, R., Ray, A., Schneider, J., Zaremba, W. and Abbeel, P. (2017). Domain randomization for transferring deep neural networks from simulation to the real world, 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Vancouver, BC, Canada, 23–30, doi: 10.1109/IROS.2017.8202133. [TH12] Tieleman, T. and Hinton, G. (2012). Lecture 6.5-rmsprop: Divide the gradient by a running average of its recent magnitude. COURSERA Neural Netw. Mach. Learn. 4, 26–31. [TP22] Tomalini, A. and Pristeri, E. (2022). Real-Time Identification of Artifacts: Synthetic Data for AI Model. In A. Giordano, M. Russo and R. Spallone (eds.), Representation Challenges: New Frontieres of AR and AI Research for Cultural Heritage and Innovative Design, FrancoAngeli, 227–34. 10.3280/oa-845-c216. + [TPAB 18] Tremblay, J., Prakash, A., Acuna, D., Brophy, M., Jampani, V., Anil, C., To, T., Cameracci, E., Boochoon, S. and Birchfield, S. (2018). Training deep networks with synthetic data: Bridging the reality gap by domain randomization. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 1082–8. [TPB21] Tomalini, A., Pristeri, E. and Bergamasco, L. (2021). Photogrammetric Survey for a Fast Construction of Synthetic Dataset. 10.3280/oa-686.34. [Tsi22] Tsirikoglou, A. (2022). Synthetic data for visual machine learning: A data-centric approach, PhD dissertation, Linköping University Electronic Press. ↗
  6. Tremblay, J., To, T., Sundaralingam, B., Xiang, Y., Fox, D. and Birchfield, S. (2018). Deep Object Pose Estimation for Semantic Robotic Grasping of Household Objects. [VDPA+21] Vögtlin, L., Drazyk, M., Pondenkandath, V., Alberti, M. and Ingold, R. (2021). Generating Synthetic Handwritten Historical Documents With OCR Constrained GANs. [VW21] Valtchev, S. Z., and Wu, J. (2021). Domain randomization for neural network classification. Journal of big data 8(1), 94. https://doi.org/10.1186/s40537-021-00455-5. [WSSP+23] Williams, E. C., Su, G., Schloen, S. R., Prosser, M. C., Paulus, S. and Krishnan, S. (2023). DeepScribe: Localization and Classification of Elamite Cuneiform Signs Via Deep Learning. ArXiv, abs/2306.01268. [WTWC21] Weir, H., Thompson, K., Woodward, A., Choi, B., Braun, A. and Martinez, T. (2021). ChemPix: Automated Recognition of Hand-drawn Hydrocarbon Structures Using Deep Learning. Chemical Science 12. 10.1039/D1SC02957F. [You00] Younger, K. L. (2000). The Hadad Inscription. In K. L. Younger, The Context of Scripture, vol. 2, 156–58. Leiden, Netherlands: Brill. [Eda13] Edan, N. M. (2013). Cuneiform symbols recognition based on k-means and neural network. AL-Rafidain Journal of Computer Sciences and Mathematics, 10(1), 195–202. [MA14] Mostofi, F. and Khashman, A. (2014). Intelligent recognition of ancient Persian Cuneiform characters. In International Conference on Neural Computation Theory and Applications, 119-123. [COG16] Can, G., Odobez, J.-M. and Gatica-Perez, D. (2016). Evaluating shape representations for Maya glyph classification. Journal on Computing and Cultural Heritage (JOCCH), 9(3), 1–26. [FMMN18] Firmani, D., Maiorino, M., Merialdo, P. and Nieddu, E. (2018). Towards knowledge discovery from the Vatican secret archives. in codice ratio-episode 1: Machine transcription of the manuscripts. In ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 263–272. [FvG13] Franken, M. and van Gemert, J. C. (2013). Automatic Egyptian hieroglyph recognition by retrieving images as texts. In ACM International Conference on Multimedia, 765–768, Association for Computing Machinery. [SCHK+21] Swindall, M. I., Croisdale, G., Hunter, C. C., Keener, B., Williams, A. C., Brusuelas, J. H., Krevans, N., Sellew, M., Fortson, L. and Wallin. J. F. (2021). Exploring learning approaches for ancient Greek character recognition with citizen science data. In International Conference on eScience, 128–137. [SPKW+22] Swindall, M. I., Player, T., Keener, B., Williams, A. C., Brusuelas, J. H., Nicolardi, F., D’Angelo, M., Vergara, C., McOsker, M. and Wallin, J. F. (2022). Dataset augmentation in papyrology with generative models: A study of synthetic ancient Greek character images. In International Joint Conference on Artificial Intelligence (IJCAI), 4973–4979. ↗
  7. Haliassos, A., Barmpoutis, P., Stathaki, T., Quirke, S., and Constantinides, A. (2020). Classification and detection of symbols in ancient papyri. In Visual Computing for Cultural Heritage. Springer, 121–140. [MHHA+22] Moustafa, R., Hesham, F., Hussein, S., Amr, B., Refaat, S., Shorim, N. and Ghanim, T. M. (2022). Hieroglyphs language translator using deep learning techniques (Scriba). In International Mobile, Intelligent, and Ubiquitous Computing Conference (MIUCC), 125–132, IEEE. ↗

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/hbo-2023-aioanei-deep-aramaic-towards-synthetic-data/