Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userWhat Have Been Learned & What Should Be Learned? An Empirical Study of How to Selectively Augment Text for Classification
63
0
0.0
(
0
)
Added by
Biyang Guo
Publication date
2021
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Text augmentation techniques are widely used in text classification problems to improve the performance of classifiers, especially in low-resource scenarios. Whilst lots of creative text augmentation methods have been designed, they augment the text in a non-selective manner, which means the less important or noisy words have the same chances to be augmented as the informative words, and thereby limits the performance of augmentation. In this work, we systematically summarize three kinds of role keywords, which have different functions for text classification, and design effective methods to extract them from the text. Based on these extracted role keywords, we propose STA (Selective Text Augmentation) to selectively augment the text, where the informative, class-indicating words are emphasized but the irrelevant or noisy words are diminished. Extensive experiments on four English and Chinese text classification benchmark datasets demonstrate that STA can substantially outperform the non-selective text augmentation methods.
rate research
Read More
Anonymous peer review is used by the great majority of computer science conferences. OpenReview is such a platform that aims to promote openness in peer review process. The paper, (meta) reviews, rebuttals, and final decisions are all released to public. We collect 5,527 submissions and their 16,853 reviews from the OpenReview platform. We also collect these submissions citation data from Google Scholar and their non-peer-review
Numerous models for grounded language understanding have been recently proposed, including (i) generic models that can be easily adapted to any given task and (ii) intuitively appealing modular models that require background knowledge to be instantiated. We compare both types of models in how much they lend themselves to a particular form of systematic generalization. Using a synthetic VQA test, we evaluate which models are capable of reasoning about all possible object pairs after training on only a small subset of them. Our findings show that the generalization of modular models is much more systematic and that it is highly sensitive to the module layout, i.e. to how exactly the modules are connected. We furthermore investigate if modular models that generalize well could be made more end-to-end by learning their layout and parametrization. We find that end-to-end methods from prior work often learn inappropriate layouts or parametrizations that do not facilitate systematic generalization. Our results suggest that, in addition to modularity, systematic generalization in language understanding may require explicit regularizers or priors.
The subject of micro-variability among Mira stars has received increased attention since DeLaverny et al. (1998) reported short-term brightness variations in 15 percent of the 250 Mira or Long Period Variable stars surveyed using the broadband 340 to 890 nm Hp filter on the HIPPARCOS satellite. The abrupt variations reported ranged 0.2 to 1.1 magnitudes, on time-scales between 2 to 100 hours, with a preponderance found nearer Mira minimum light phases. However, the HIPPARCOS sampling frequency was extremely sparse and required confirmation because of potentially important atmospheric dynamics and dust-formation physics that could be revealed. We report on Mira light curve sub-structure based on new CCD V and R band data, augmenting the known light curves of Hipparcos-selected long period variables [LPVs], and interpret same in terms of [1] interior structure, [2] atmospheric structure change, and/or [3] formation of circumstellar [CS] structure. We propose that the alleged micro-variability among Miras is largely undersampled, transient overtone pulsation structure in the light curves.
As the success of deep models has led to their deployment in all areas of computer vision, it is increasingly important to understand how these representations work and what they are capturing. In this paper, we shed light on deep spatiotemporal representations by visualizing what two-stream models have learned in order to recognize actions in video. We show that local detectors for appearance and motion objects arise to form distributed representations for recognizing human actions. Key observations include the following. First, cross-stream fusion enables the learning of true spatiotemporal features rather than simply separate appearance and motion features. Second, the networks can learn local representations that are highly class specific, but also generic representations that can serve a range of classes. Third, throughout the hierarchy of the network, features become more abstract and show increasing invariance to aspects of the data that are unimportant to desired distinctions (e.g. motion patterns across various speeds). Fourth, visualizations can be used not only to shed light on learned representations, but also to reveal idiosyncracies of training data and to explain failure cases of the system.
I revisit two theories of cell differentiation in multicellular organisms published a half-century ago, Stuart Kauffmans global gene regulatory dynamics (GGRD) model and Roy Brittens and Eric Davidsons modular gene regulatory network (MGRN) model, in light of newer knowledge of mechanisms of gene regulation in the metazoans (animals). The two models continue to inform hypotheses and computational studies of differentiation of lineage-adjacent cell types. However, their shared notion (based on bacterial regulatory systems) of gene switches and networks built from them, have constrained progress in understanding the dynamics and evolution of differentiation. Recent work has described unique write-read-rewrite chromatin-based expression encoding in eukaryotes, as well metazoan-specific processes of gene activation and silencing in condensed-phase, enhancer-recruiting regulatory hubs, employing disordered proteins, including transcription factors, with context-dependent identities. These findings suggest an evolutionary scenario in which the origination of differentiation in animals, rather than depending exclusively on adaptive natural selection, emerged as a consequence of a type of multicellularity in which the novel metazoan gene regulatory apparatus was readily mobilized to amplify and exaggerate inherent cell functions of unicellular ancestors. The plausibility of this hypothesis is illustrated by the evolution of the developmental role of Grainyhead-like in the formation of epithelium.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
