Transductive Learning with String Kernels for Cross-Domain Text Classification

Ionescu, Radu Tudor; Butnaru, Andrei M.

doi:10.1007/978-3-030-04182-3_42

Cited by 6 publications

(6 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several works that followed (e.g., [3,44]) adopted this definition of "transductive transfer learning". To the best of our knowledge, the only works about transfer learning which use the term "transduction" in Vapnik's original sense are [46,47] (although they presents no text classification experiments) and [2,23] (which will be discussed in Section 4.2).…”

Section: The Shifting Meaning Of "Transduction"mentioning

confidence: 99%

“…Examples of NMF techniques include Topical Correspondence Transfer (TCT) [63] for cross-domain adaptation, Semi-supervised Matrix Completion (SSMC) [57], Two-Step Learning (TSL) [56], and the Subspace Learning Framework (CL-SLF) [62] for cross-lingual adaptation. Very recently, [23] has proposed TKC, a transductive method based on string kernels that was also evaluated on the MDS dataset.…”

Section: Inductive Transfer Problemsmentioning

confidence: 99%

“…This means that, even assuming the approaches could have been trained on a different set of parallel documents (and this is something which incidentally remains unclear), the truth is that the results they reported are inevitably optimized for the specific test documents (unfairly taken to be the object set), and can thus not be granted to be representative of the more general inductive transfer problem. TransDANN [2] and TKC [23] also fall in the "transductive group", though in these cases the incursion was deliberated and openly acknowledged.…”

Section: Inductive Transfer Problemsmentioning

confidence: 99%

“…22 Apart from bypassing the translation of pivot words when the source and target languages are the same, and apart from implementing the normalised concatenation described in Section 5.1, the main change we have made concerns the replacement of the original learning device in charge of the final predictions with SVM liдht . 23 TSCL is thus obtained by having SVM l iдht operate in transductive mode and making the object set (with labels omitted) available at training time. In previous literature, SCL has been tested on Webis-CLS-10 [43] and on MDS [7].…”

Section: A =mentioning

confidence: 99%

See 3 more Smart Citations

Lost in Transduction: Transductive Transfer Learning in Text Classification

Moreo

Esuli

Sebastiani

2021

ACM Trans. Knowl. Discov. Data

View full text Add to dashboard Cite

Obtaining high-quality labelled data for training a classifier in a new application domain is often costly. Transfer Learning (a.k.a. “Inductive Transfer”) tries to alleviate these costs by transferring, to the “target” domain of interest, knowledge available from a different “source” domain. In transfer learning the lack of labelled information from the target domain is compensated by the availability at training time of a set of unlabelled examples from the target distribution. Transductive Transfer Learning denotes the transfer learning setting in which the only set of target documents that we are interested in classifying is known and available at training time. Although this definition is indeed in line with Vapnik’s original definition of “transduction”, current terminology in the field is confused. In this article, we discuss how the term “transduction” has been misused in the transfer learning literature, and propose a clarification consistent with the original characterization of this term given by Vapnik. We go on to observe that the above terminology misuse has brought about misleading experimental comparisons, with inductive transfer learning methods that have been incorrectly compared with transductive transfer learning methods. We then, give empirical evidence that the difference in performance between the inductive version and the transductive version of a transfer learning method can indeed be statistically significant (i.e., that knowing at training time the only data one needs to classify indeed gives an advantage). Our clarification allows a reassessment of the field, and of the relative merits of the major, state-of-the-art algorithms for transfer learning in text classification.

show abstract

Section: The Shifting Meaning Of "Transduction"mentioning

confidence: 99%

Section: Inductive Transfer Problemsmentioning

confidence: 99%

Section: Inductive Transfer Problemsmentioning

confidence: 99%

Section: A =mentioning

confidence: 99%

See 2 more Smart Citations

Lost in Transduction: Transductive Transfer Learning in Text Classification

Moreo

Esuli

Sebastiani

2021

ACM Trans. Knowl. Discov. Data

View full text Add to dashboard Cite

show abstract

“…Transductive learning approaches actively take advantage of sample distributions during inference. This paradigm has been used to improve performance on tasks such as crossdomain text classification (Ionescu and Butnaru 2018) and neural machine translation (Poncelas, de Buy Wenniger, and Way 2019). However, these studies do not address cases where the original dataset lacks enough labeled examples.…”

Section: Related Workmentioning

confidence: 99%

Domain-agnostic Document Representation Learning Using Latent Topics and Metadata

Raman¹,

Nourbakhsh²,

Shah³

et al. 2021

FLAIRS

View full text Add to dashboard Cite

Fine-tuning a pre-trained neural language model with a task specific output layer is the de facto approach of late when dealing with document classification. This technique is inadequate when labeled examples are unavailable at training time and when the metadata artifacts in a document must be exploited. We address these challenges by generating document representations that capture both text and metadata in a task agnostic manner. Instead of traditional auto-regressive or auto-encoding based training, our novel self-supervised approach learns a soft-partition of the input space when generating text embeddings by employing a pre-learned topic model distribution as surrogate labels. Our solution also incorporates metadata explicitly rather than just augmenting them with text. The generated document embeddings exhibit compositional characteristics and are directly used by downstream classification tasks to create decision boundaries from a small number of labels, thereby eschewing complicated recognition methods. We demonstrate through extensive evaluation that our proposed cross-model fusion solution outperforms several competitive baselines on multiple domains.

show abstract