Nöral Makine Çevirisinde Plastisite: Bilişsel ve Teknik bir Perspektif

Öz

Nöroplastisite, insan beyninin değişken koşullara adapte olması ve uyum sağlaması konusundaki esnekliği ifade eden bir kavramdır. Yapay zekâ sistemlerinde ve nöral makine çevirisinde ise plastisite, çeviri çıktılarının doğruluğu ve bağlama duyarlılığı ile ilgili bir konudur. Bu çalışma, nöral makine çevirisinde plastisitenin nasıl sağlandığını ve bunu sağlayan mekanizmaların çeviri sürecindeki etkisini incelemeyi amaçlamaktadır. Plastisite olgusu, çalışmada veri temelli kavramsal analiz yaklaşımıyla ele alınmış, teknik boyutunun yanı sıra erek dilde anlamın yeniden yapılandırılması ve bağlamsal esnekliğin işlevselliği bakımından incelenmiştir. Bu doğrultuda dil modellerinde plastisiteyi sağlayan iki temel mekanizma ele alınmıştır: Birincisi hataya dayalı öğrenme süreci olan ve geri yayılım anlamına gelen backpropagation, ikincisi ise bir dil modelinin kullanıcı tarafından yeni girdilere uyumlanmasını içeren fine-tuning işlemi. Bu işleyişin arkasındaki mekanizma ise attention terimiyle ifade edilmektedir. Bu mekanizmalar öncelikle teknik özellikleriyle alanyazında araştırılmış, makine öğrenmesindeki plastisitenin insan beynindeki plastisite kavramından farkı ifade edilmiştir. Ardından bu mekanizmaların işlevselliği, örnek metinler, dikkat hari taları ve eğitilen bir dil modeli üzerinden yürütülen doküman analizi kapsamında betimleyici bir biçimde analiz edilmiştir. Elde edilen bulgular çeviribilimsel bakış açısından bilişsel süreçlerle ilişkilendirilmiş, model çıktılarının bağlamsal esneklik kapasitesi incelenmiştir. Bulgular, yapay öğrenme sistemlerinin nöral makine çevirisi yaparken yüzeysel anlam eşleştirmede başarılı olduğunu, bununla birlikte modelin işleyişinde istenmeyen yan etkilere de yol açabileceğini göstermiştir.

Anahtar Kelimeler

• dikkat
• ince ayar
• geri yayılım
• nöral makine çevirisi
• plastisite

Plasticity in Neural Machine Translation: A Cognitive and Technical Perspective

Öz

Neuroplasticity refers to the flexibility of the human brain in adapting to changing conditions. In artificial intelligence systems and neural machine translation (NMT), plasticity is related to the accuracy and contextual sensitivity of the translation outputs. This study aims to explore how plasticity is achieved in NMT and how its underlying mechanisms affect the translation process. The phenomenon is addressed through a data-driven conceptual analysis, focusing not only on its technical dimension but also on meaning reconstruction in the target language and the functional role of contextual flexibility. In this regard, two core mechanisms enabling plasticity are examined: backpropagation, an error-based learning process, and fine-tuning, which involves adapting a language model to new input by users. The underlying operation is driven by attention mechanisms. These processes are f irst investigated through the technical literature, highlighting how machine learning plasticity differs from human neuroplasticity. Then, their functionality is analyzed descriptively through a document analysis involving sample texts, attention maps, and an adapted language model. The findings are associated with cognitive processes from a translation studies perspective, showing that while NMT performs well in surface-level meaning matching, it may also produce unintended side effects within its operational logic.

Anahtar Kelimeler

• attention
• backpropagation
• fine-tuning
• neural machine translation
• plasticity

Kaynakça

1. Abbas, Z., Zhao, R., Modayil, J., White, A. & Machado, M. C. (2023). Loss of plasticity in continual deep reinforcement learning. In Conference on lifelong learning agents, 232: 620-636. PMLR. Retrieved from: https://proceedings.mlr.press/v232/abbas23a.html google scholar
2. Açıkgöz, N. & Madi, B. (2013). Öğrenme ile Beyinde Oluşan Değişiklikler (Plastisite). Marmara Üniversitesi Atatürk Eğitim Fakültesi Eğitim Bilimleri Dergisi, 9(9), 29-36. google scholar
3. Allemann, A., Atrio, À. R. & Popescu-Belis, A. (2024). Optimizing the Training Schedule of Multilingual NMT using Reinforcement Learning. google scholar
4. Preprint arXiv, Computation and Language. https://doi.org/10.48550/arXiv.2410.06118 google scholar
5. Anlar, B. (2013). Beyinde Plastisite ve Bozuklukları. Türkiye Klinikleri Pediatrik Bilimler- Özel Konular, 9(4), 129-137. google scholar
6. Aslan, E. (2024). Yapay Zekâ Destekli Çeviri Araçlarının Edebi Çevirideki Yeterlilikleri Üzerine Karşılaştırmalı Bir İnceleme. IU Journal of Translation Studies, (20), 32-45. google scholar
7. Bahdanau, D., Cho, KH. & Bengio, Y. (2015). Neural machine translation by jointly learning to align and translate. In Proceedings of ICLR. Preprint arXiv, https://doi.org/10.48550/arXiv.1409.0473 google scholar
8. Baker, M. (1992). In Other Words: A Coursebook on Translation. Routledge, London and New York. Retrieved from: https://archive.org/details/InOtherWordsByMonaBaker/page/n3/mode/2up google scholar

9. Bourdieu, P. (2023). Genel Sosyoloji: Collège de France Dersleri (1981-1983) (Z. Emirosmanoğlu, Çev.), İletişim Yayınları. google scholar
10. Chen, N. (2024). Text Classification Model Based on Long Short-Term Memory with L2 Regularization. In 2024 Second International Conference on Data Science and Information System (ICDSIS), 1-4. IEEE. Doi: 10.1109/ICDSIS61070.2024.10594621 google scholar
11. Retrieved from: https://www.researchgate.net/publication/382377464_Text_Classification_Model_Based_on_Long_Short-Term_Memory_with_L2_Regularization google scholar
12. Corbetta, M. & Shulman, G. L. (2002). Control of goal-directed and stimulus-driven attention in the brain. Nature reviews neuroscience, 3(3), 201-215. Doi: 10.1038/nrn755 https://www.researchgate.net/publication/11375373_Control_of_Goal-Directed_and_Stimulus-Driven_Attention_in_the_Brain google scholar
13. Cüceloğlu, D. (1995). İnsan ve Davranışı: Psikolojinin Temel Kavramları, 25. Basım Remzi Kitapevi. google scholar
14. Çetin, Ö., & Duran, A. (2024). A Comparative Analysis Of The Performances of ChatGPT, DeepL, Google Translate And A Human Translator In Community Based Settings. Amasya Üniversitesi Sosyal Bilimler Dergisi, 9(15), 120-173. google scholar
15. DWDS Digitales Wörterbuch der Deutschen Sprache. (2025, 24 Mayıs). https://www.dwds.de/wb/Verfahren google scholar
16. Dohare, S., Fernando Hernandez-Garcia, J., Lan, Q., Rahman, P., Rupam Mahmood, A. & S.Sutton, R. (2024). Loss of plasticity in deep continual learning. Nature, Springer Science and Business Media LLC. https://doi.org/10.1038/s41586-024-07711-7 google scholar
17. French, R. M. (2003). Catastrophic interference in connectionist networks. Nadel L. (Ed.) In Encyclopedia of Cognitive Sciences (Vol.1, s. 431-435). London: Nature Publishing Group. Retrieved from: http://leadserv.u-bourgogne.fr/rfrench/french/catastrophic_forgetting.ECS.french.pdf google scholar
18. Gao, Y., Xiong, Y., Gao, X., Jia, K., Pan, J., Bi, Y., … & Wang, H. (2023). Retrieval-augmented generation for large language models: A survey. arXiv preprint arXiv:2312.10997, 2(1). google scholar
19. Genç, A. & Çınar Yağcı, Ş. (2024). Çeviribilim Alanında Yapay Zekâ Üzerine Ulusal Alan Yazında Yazılmış Makalelerin Eğilimleri Üzerine Bir Araştırma. Kırıkkale Üniversitesi Sosyal Bilimler Dergisi, 14(3), 119^136. google scholar
20. Gu, S. & Feng, Y. (2020). Investigating catastrophic forgetting during continual training for neural machine translation. In Proceedings of the 28th International Conference on Computational Linguistics, 4315–4326, Barcelona, Spain (Online). International Committee on Computational Linguistics. Doi: 10.18653/v1/2020.coling^main.381 Retrieved from: https://aclanthology.org/2020.coling^main.381.pdf google scholar
21. Hatim, B. & Mason, I. (2005). The Translator as Communicator. Taylor & Francis e-Library. First pub. (1997) Routledge, London and New York. ISBN 0-203-99272-5 google scholar
22. Herholz, K., Langen, K. J., Schiepers, C. & Mountz, J. M. (2012). Brain tumors. In Seminars in nuclear medicine 42 (6), 356-370. WB Saunders. Doi:10.1053/j.semnuclmed.2012.06.001. google scholar
23. Hinke, R. M., Hu, X., Stillman, A. E., Kim, S.^g., Merkle, H., Salmi, R. & Ugurbil, K. (1993). Functional magnetic resonance imaging of Broca's area during internal speech. Neuroreport: An International Journal for the Rapid Communication of Research in Neuroscience, 4(6), 675–678. https://doi.org/10.1097/00001756^199306000^00018 google scholar
24. Jääskeläinen, R. (2011). Focus on methodology in think-aloud studies on translating. In Tapping and mapping the processes of translation and interpreting: Outlooks on empirical research (s. 71-82). John Benjamins Publishing Company. https://doi.org/10.1075/btl.37.08jaa google scholar
25. Kerimoğlu, C. (2022). Dilin Kökeni Arayışları-5: Beyin ve Dil. Dil Araştırmaları, 16(30), 21-37. https://doi.org/10.54316/dilarastirmalari.1075944 google scholar
26. Kılıç, T. (2024). Yeni Bilim: Bağlantısallık Yeni Kültür: Yaşamdaşlık, 7. Basım, Ayrıntı Yayınları, İstanbul. Birinci Basım 2021. google scholar
27. Kocmi, T. (2020). Exploring Benefits of Transfer Learning in Neural Machine Translation. (Doctoral dissertation, Institute of Formal and Applied Linguistics, Charles University). Retrieved from: https://dspace.cuni.cz/bitstream/handle/20.500.11956/115854/140081447.pdf?sequence=1 google scholar
28. Kumlu, D., & Okul, M. (2023). Yapay Zekâ ile Çeviri Uygulamaları: Kültürel Unsurların Çevirisinde ChatGPT. Contemporary Translation Studies, 24. google scholar
29. Liang, J., Zhao, Ch., Wang, M., Qui, L. & Li, L. (2021). Finding Sparse Structures for Domain Specific Neural Machine Translation. The Thirty-Fifth AAAI Conference on Artificial Intelligence (AAAI-21). Retrieved from: https://cdn.aaai.org/ojs/17574/17574-13-21068-1-2-20210518.pdf google scholar
30. Nord, C. (2014). Translating as a Purposeful Activiy: Functionalist Approaches Explained. Routledge, London and New York. First pub. (1997), St. Jerome Publishing. google scholar
31. Odacıoğlu, M. C. (2024). Yapay Zekâ ve İnsan Çevirisi: Hukuk Metinlerine Dayalı Karşılaştırmalı Bir Çalışma. Karamanoğlu Mehmetbey Üniversitesi Uluslararası Filoloji ve Çeviribilim Dergisi, 6(1), 147-171. https://doi.org/10.55036/ufced.1477008 google scholar
32. PACTE GROUP. (2017). Decision-making. In Researching Translation Competence by PACTE Group (191-210). John Benjamins Publishing Company. google scholar
33. Posner, M. I. & Petersen, S. E. (1990). The attention system of the human brain. Annual review of neuroscience, 13(1), 25-42. Doi: 10.1146/annurev.ne.13.030190.000325 Retrieved from: https://www.researchgate.net/publication/20971732_The_Attention_System_of_the_Human_Brain google scholar
34. Price, C. J., Green, D. W. & Von Studnitz, R. (1999). A functional imaging study of translation and language switching. Brain, 122(12), 2221-2235. google scholar
35. Reiß, K. & Vermeer, H. (1984). Grundlegung einer allgemeinen Translationstheorie. De Gruyter. google scholar
36. Rumelhart, D. E., Hinton, G.E. & Williams, R.J. (1986). Learning representations by back-propagating errors. Nature, 323(6088), 533-536. google scholar
37. Şahin, M. (2023). Yapay Çeviri. Çeviribilim Yayınları. google scholar
38. Schlegel, P., Yin, Y., Bates, A. S., Dorkenwald, S., Eichler, K., Brooks, P., … Jefferis, G. S. (2024). Whole-brain annotation and multi- connectome cell typing of Drosophila. Nature, 634(8032), 139-152. https://doi.org/10.1038/s41586-024-07686-5 google scholar
39. Snelleman, E. (2016). Decoding neural machine translation using gradient descent (Master Thesis in Computer Science, Algorithms, Language and Logic, Department of Computer Science and Engineering, Chalmers University of Technology, Gothenburg). Retri- eved from: https://odr.chalmers.se/server/api/core/bitstreams/c2b7d271-c5f9-472c-b0b7-36bd8eef1d40/content google scholar
40. Stanford Online (2016, 27 Nisan). Can the brain do back-propagation? Geoffrey Hinton. [Video]. YouTube. https://www.youtube.com/watch?v=VIRCybGgHts google scholar
41. Teke Tek Bilim (2024, 26 Mayıs). İnsan beyninin sınırları? Prof. Dr. Türker Kılıç & Fatih Altaylı. [Video]. YouTube. https://www.youtube.com/watch?v=jUZvlU1It5U google scholar
42. Thompson, B., Gwinnup, J., Khayrallah, H., Duh, K. & Koehn, P. (2019). Overcoming catastrophic forgetting during domain adaptation of neural machine translation. In 2019 Annual Conference of the North American Chapter of the Association for Computational Linguistics, 2062-2068. Retrieved from: https://aclanthology.org/N19^1209.pdf google scholar
43. Variš, D. & Bojar, O. (2019). Unsupervised pretraining for neural machine translation using elastic weight consolidation. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics: Student Research Workshop, 130-135. Florence, Italy. Retrieved from: https://aclanthology.org/P19-2017.pdf google scholar
44. Vaswani, A., Shazeer, N., Parmar, N., Uskoreit, J., Jones, L., Gomez, A. N., Kaiser. L. & Polosukhin, I. (2017). Attention Is All You Need. In 31st Conference on Neural Information Processing Systems, 5998-6008. Long Beach, CA. https://doi.org/10.48550/arXiv.1706.03762 google scholar
45. Vig, J. (2019). A multiscale visualization of attention in the transformer model”. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics: System Demonstrations, 37–42, Florence, Italy. Preprint arXiv https://doi.org/10.48550/arXiv.1906.05714 google scholar
46. Vrbančič, G. & Podgorelec, V. (2020). Transfer learning with adaptive fine-tuning. IEEE Access, 8, 196197-196211.Retrieved from: https://www.researchgate.net/publication/345713638_Transfer_Learning_With_Adaptive_Fine-Tuning google scholar
47. Wenger, E. & Lövdén, M. (2016). The learning hippocampus: Education and experience‐dependent plasticity. Mind, Brain, and Education, 10(3), 171-183. https://doi.org/10.1111/mbe.12112 google scholar
48. Wu, J., Liu Y. & Zong C. (2024). F-MALLOC: Feed-forward Memory Allocation for Continual Learning in Neural Machine Translation. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 7180–7192, Mexico. Association for Computational Linguistics. Main conference of NAACL. Retrieved from: https://aclanthology.org/2024.naacl-long.398/ google scholar
49. Zhao, Y., Zhang, J. & Zong, C. (2023). Transformer: A general framework from machine translation to others. Machine Intelligence Research, 20(4), 514-538. Doi: 10.1007/s11633-022-1393-5 Retrieved from: https://nlpr.ia.ac.cn/cip/ZongPublications/2023/2023-ZhaoYang-MIR.pdf google scholar