Last Words: Computational Linguistics and Deep Learning

By Christopher D. Manning. 2015

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

What do we need to build explainable AI systems for the medical domain? (2017, background): “As opposed to ﬁelds such as speech or image processing, the improvements recently gained with deep learning on text are rather modest, yet its use is very attractive since neural representations reduce the workload of manually crafting features enormously (Manning, 2015).”
Survey of the State of the Art in Natural Language Generation: Core tasks, applications and evaluation (2017, background): “In a recent opinion piece, Manning (2015) suggested that the contribution of deep learning to nlp has to date been mainly due to the power of distributed representations, rather than the exploitation of the ‘depth’ of multi-layered models.”
Is neural language acquisition similar to natural? A chronological probing study (2022, methodology): “The role of deep learning language models has been increasing in the ﬁeld of methodology for linguistic research, providing new methods for both diachronic and synchronic studies (Manning, 2015).”
Deep Active Learning for Named Entity Recognition (2017, background): “Deep active learning While DNNs have achieved impressive empirical results across diverse applications (Krizhevsky et al., 2012; Hinton et al., 2012; Manning, 2016), active learning approaches for these models have yet to be well studied, and most current work addresses image classification.”
Diachronic word embeddings and semantic shifts: a survey (2018, background): “The release of the Google Books Ngrams corpus1 played an important role in the development of the field and spurred work on the new discipline of ‘culturomics,’ studying human culture through digital media (Michel et al., 2011). Mihalcea and Nastase (2012) used this dataset to detect differences in ”
Computational Sociolinguistics: A Survey (2016, background): “ of the work within CL is centered around highly speciﬁc analysis tasks that are isolated from scenarios of use, and the focus on the obtained performance ﬁgures for such tasks is fairly dominant. As Manning (2015) mentions: "[..], there has been an overfocus on numbers, on beating the state of”
Adversarial Initialization - when your network performs the way I want (2019, background): “The usage ranges from computer vision [13] over autonomous driving, natural language processing [23] and Malware detection [32] to healthcare [25].”
Strong Baselines for Neural Semi-Supervised Learning under Domain Shift (2018, background): “Recent NLP conferences witnessed a “tsunami” of deep learning papers (Manning, 2015), followed by what we call a multi-task learning “wave”: MTL has been successfully applied to a wide range of NLP tasks (Cohn and Specia, 2013; Cheng et al.”
Towards textual data augmentation for neural networks: synonyms and maximum loss (2019, background): “One of the most recent neural network architectures that is specially designed for NLP tasks is the Transformer.”