পদ্ধতির তুলনা করুন

	কন্ট্রাস্টিভ লার্নিং ফর এনএলপি	স্ব-পর্যবেক্ষণাধীন শিখন
ক্ষেত্র	টেক্সট খনন	যন্ত্র শিখন
পরিবার	Process / pipeline	Machine learning
উদ্ভবের বছর	2020–2021	2018–2020
প্রবর্তক	Gao, Yao & Chen (SimCSE, 2021); Khosla et al. (Supervised Contrastive, 2020)	LeCun, Y. and community (formalized ~2018–2020)
ধরন	Self-supervised / supervised representation learning	Representation learning paradigm
মৌলিক উৎস	Gao, T., Yao, X., & Chen, D. (2021). SimCSE: Simple Contrastive Learning of Sentence Embeddings. Proceedings of EMNLP 2021.	LeCun, Y. & Misra, I. (2022). Self-supervised learning: The dark matter of intelligence. Meta AI Blog. https://ai.facebook.com/blog/self-supervised-learning-the-dark-matter-of-intelligence/
অপর নাম	SimCSE, contrastive sentence embeddings, ContrastiveBERT, Karşıtlık Öğrenmesi — NLP (Contrastive Learning)	SSL, self-supervised pre-training, pretext-task learning, unsupervised representation learning
সম্পর্কিত	4	3
সারসংক্ষেপ	Contrastive learning for NLP is a representation-learning technique — popularised by SimCSE (Gao et al., 2021) and Supervised Contrastive Learning (Khosla et al., 2020) — that trains a text encoder by pulling embeddings of similar text pairs together while pushing embeddings of dissimilar pairs apart. The result is a dense, high-quality embedding space that can be learned with no labels at all, or with minimal supervision, making it especially valuable when annotated data are scarce.	Self-supervised learning (SSL) is a machine-learning paradigm that generates its own supervisory signal directly from unlabeled data by defining an auxiliary pretext task — such as predicting masked words, rotating images, or contrasting augmented views — and uses the learned representations as a powerful starting point for downstream tasks with minimal labeled examples.