Low-light images often suffer from significant noise and detail loss, making it challenging to effectively distinguish signals from noise when processed directly in the spatial domain. To this end, we introduce frequency domain information to better distinguish high-frequency details from low-frequency components, thereby suppressing noise while enhancing details. The proposed method converts sRGB images into raw-RGB images by reversing the Image Signal Processor (ISP) pipeline to avoid unnecessary effects. A frequency information interaction processing unit is then designed to enhance low-frequency textures using high-frequency information and correct high-frequency data through low-frequency information to reduce structural deformation during restoration. The signal-to-noise ratio prior is introduced in the hidden layer to further remove noise. Experimental results show that the proposed method outperforms existing color and structure enhancement methods.

Considerable attention has been devoted recently to the problem of exploring and distilling factual knowledge from Pre-trained Language Models (PLMs). While continuous prompts are currently considered the best approach, determining the number and positions of continuous prompt vectors remains challenging to obtain the optimal prompts. In this paper, we introduce a new similarity-based algorithm that automatically searches for the best positions and number of continuous prompt vectors based on the similarity score between prompt representation and the relation representation for the final optimal prompts. The prompt optimization process is done before model training, making it efficient and less time-consuming. Our experiments demonstrate that our method significantly outperforms state-of-the-art methods in probing factual knowledge.

Concepts play a central role in many applications.This includes settings where concepts have to be modelled in the absence of sentence context.Previous work has therefore focused on distilling decontextualised concept embeddings from language models.But concepts can be modelled from different perspectives, whereas concept embeddings typically mostly capture taxonomic structure.To address this issue, we propose a strategy for identifying what different concepts, from a potentially large concept vocabulary, have in common with others.We then represent concepts in terms of the properties they share with the other concepts.To demonstrate the practical usefulness of this way of modelling concepts, we consider the task of ultra-fine entity typing, which is a challenging multi-label classification problem.We show that by augmenting the label set with shared properties, we can improve the performance of the state-of-the-art models for this task. 1

Learning vectors that capture the meaning of concepts remains a fundamental challenge. Somewhat surprisingly, perhaps, pre-trained language models have thus far only enabled modest improvements to the quality of such concept embeddings. Current strategies for using language models typically represent a concept by averaging the contextualised representations of its mentions in some corpus. This is potentially sub-optimal for at least two reasons. First, contextualised word vectors have an unusual geometry, which hampers downstream tasks. Second, concept embeddings should capture the semantic properties of concepts, whereas contextualised word vectors are also affected by other factors. To address these issues, we propose two contrastive learning strategies, based on the view that whenever two sentences reveal similar properties, the corresponding contextualised vectors should also be similar. One strategy is fully unsupervised, estimating the properties which are expressed in a sentence from the neighbourhood structure of the contextualised word embeddings. The second strategy instead relies on a distant supervision signal from ConceptNet. Our experimental results show that the resulting vectors substantially outperform existing concept embeddings in predicting the semantic properties of concepts, with the ConceptNet-based strategy achieving the best results. These findings are furthermore confirmed in a clustering task and in the downstream task of ontology completion.

While the success of pre-trained language models has largely eliminated the need for high-quality static word vectors in many NLP applications, static word vectors continue to play an important role in tasks where word meaning needs to be modelled in the absence of linguistic context. In this paper, we explore how the contextualised embeddings predicted by BERT can be used to produce high-quality word vectors for such domains, in particular related to knowledge base completion, where our focus is on capturing the semantic properties of nouns. We find that a simple strategy of averaging the contextualised embeddings of masked word mentions leads to vectors that outperform the static word vectors learned by BERT, as well as those from standard word embedding models, in property induction tasks. We notice in particular that masking target words is critical to achieve this strong performance, as the resulting vectors focus less on idiosyncratic properties and more on general semantic properties. Inspired by this view, we propose a filtering strategy which is aimed at removing the most idiosyncratic mention vectors, allowing us to obtain further performance gains in property induction.

Emotion-cause pair extraction is a recently proposed task that aims at extracting all potential clause-level pairs of emotion and cause in text. To solve this task, researchers first proposed a two-step pipeline method. This method extracts the emotions and causes individually in the first step, then pairs the extracted emotions and causes and filters the invalid emotion-cause pairs in the second step. Due to that the two-step method has the error accumulation problem and is hard to be optimized jointly, several one-step end-to-end models have been proposed. These models share a similar underlying idea, that is, reframing the emotion-cause pair extraction task as a classification problem of candidate clause pairs. Unlike these models, in this paper, we reframe the emotion-cause pair extraction task as a unified sequence labeling problem, which allows to extract emotion-cause pairs through one pass of sequence labeling. This is realized by designing a special set of unified labels. In the unified label, we design a content part for emotion/cause identification and a pairing part for clause pairing. Then the emotion-cause pairs can be implicitly derived from the unified labels. To address this unified sequence labeling problem, we propose a unified target-oriented sequence-to-sequence model, which comprehensively utilizes the information of target clause, global context, and former decoded label, to perform end-to-end unified sequence labeling. The experimental results demonstrate the effectiveness of both our proposed unified sequence labeling scheme and unified target-oriented sequence-to-sequence model. All the code and data of this work can be obtained at https://github.com/zifengcheng/UTOS.

Grain segmentation of sandstone images is to partition the images into non‐overlapping regions, each of which is an independent mineral grain. However, a sandstone image usually contains hundreds of mineral grains and complicated micro‐structures, which makes current segmentation methods inefficient. In this study, the authors propose a three‐stage framework for the automatic segmentation of sandstone images. In the first stage, the input sandstone images are pre‐segmented into over‐segmented mineral superpixels. In the second stage, the instance‐independent features are extracted by a specially designed convolutional neural network, and the instance‐aware features are extracted by computing histogram statistics and Gabor responses of mineral superpixels. In the third stage, a novel weighted fuzzy clustering algorithm is proposed to cluster the mineral superpixels into different classes, afterwards the adjacent mineral superpixels are merged to yield the complete minerals according to their classes. The experimental results conducted on the sandstone image datasets demonstrate the effectiveness of the proposed method, which evidently outperform the state‐of‐the‐art segmentation methods.

Grain segmentation of sandstone images is to segment the mineral grains into separate regions, which is the first step for mineral identification and sandstone classification. A sandstone image contains a large number of mineral grains, which due to the complexity of the micro-structures and the variability of the optical properties, becomes difficult for automatic segmentation. In this paper, we propose a three-stage method for grain segmentation taking multi-angle sandstone images as the input. In the first stage, the pixels of the images are clustered by both color and spatial properties into superpixels. In the second stage, convolutional neural network (CNN) is trained based on replicated images of sandstone mineral grains, which is then used to extract the convolutional features of the superpixels. In the third stage, the fuzzy clustering algorithm is used to merge the over-segmented superpixels into mineral grains. We collect sandstone images from Tibet, based on which the experimental results demonstrate the following: (1) The convolutional features extracted by the designed CNN is more suitable for characterizing the mineral grains of sandstone images, comparing to the handcrafted features. (2) The proposed three-stage method is more effective than the state-of-the-art segmentation methods.

Grain segmentation of petrographic thin section microscopic (TSM) images is the first step for computer aided mineral identification and rock naming. The TSM images contain a large number of mineral grains and the differences among adjacent grains are usually ambiguous, which makes current segmentation technologies inefficient. In this paper, we take advantage of multi-angle TSM images and propose a method for grain segmentation. Accordingly, the method consists of two steps, in the first step, we enhance the SLIC algorithm to handle multi-angle images and produce the initial superpixels. In the second step, multiple features are extracted for comprehensive description of the superpixels, and dissimilarities between superpixels are measured according to the extracted features. Then the multi-angle region merging algorithm is employed to merge similar adjacent superpixels and get the final segmentation results. Experimental results demonstrate both the effectiveness and potential of the proposed method for grain segmentation of TSM images.