arXiv subjects

Yue Deng

Publications and source records attributed to Yue Deng.

At least 73 records · Page 4Linked to original sources

arXiv2023

SOUL: Towards Sentiment and Opinion Understanding of Language

Sentiment analysis is a well-established natural language processing task, with sentiment polarity classification being one of its most popular and representative tasks. However, despite the success of pre-trained language models in this area, they often fall short of capturing the broader complexities of sentiment analysis. To address this issue, we propose a new task called Sentiment and Opinion Understanding of Language (SOUL). SOUL aims to evaluate sentiment understanding through two subtasks: Review Comprehension (RC) and Justification Generation (JG). RC seeks to validate statements that focus on subjective information based on a review text, while JG requires models to provide explanations for their sentiment predictions. To enable comprehensive evaluation, we annotate a new dataset comprising 15,028 statements from 3,638 reviews. Experimental results indicate that SOUL is a challenging task for both small and large language models, with a performance gap of up to 27% when compared to human performance. Furthermore, evaluations conducted with both human experts and GPT-4 highlight the limitations of the small language model in generating reasoning-based justifications. These findings underscore the challenging nature of the SOUL task for existing models, emphasizing the need for further advancements in sentiment analysis to address its complexities. The new dataset and code are available at https://github.com/DAMO-NLP-SG/SOUL.

cs.CL↗

arXiv2023

Sentiment Analysis in the Era of Large Language Models: A Reality Check

Sentiment analysis (SA) has been a long-standing research area in natural language processing. It can offer rich insights into human sentiments and opinions and has thus seen considerable interest from both academia and industry. With the advent of large language models (LLMs) such as ChatGPT, there is a great potential for their employment on SA problems. However, the extent to which existing LLMs can be leveraged for different sentiment analysis tasks remains unclear. This paper aims to provide a comprehensive investigation into the capabilities of LLMs in performing various sentiment analysis tasks, from conventional sentiment classification to aspect-based sentiment analysis and multifaceted analysis of subjective texts. We evaluate performance across 13 tasks on 26 datasets and compare the results against small language models (SLMs) trained on domain-specific datasets. Our study reveals that while LLMs demonstrate satisfactory performance in simpler tasks, they lag behind in more complex tasks requiring deeper understanding or structured sentiment information. However, LLMs significantly outperform SLMs in few-shot learning settings, suggesting their potential when annotation resources are limited. We also highlight the limitations of current evaluation practices in assessing LLMs' SA abilities and propose a novel benchmark, \textsc{SentiEval}, for a more comprehensive and realistic evaluation. Data and code during our investigations are available at \url{https://github.com/DAMO-NLP-SG/LLM-Sentiment}.

cs.CL↗

arXiv2023

Bidirectional Generative Framework for Cross-domain Aspect-based Sentiment Analysis

Cross-domain aspect-based sentiment analysis (ABSA) aims to perform various fine-grained sentiment analysis tasks on a target domain by transferring knowledge from a source domain. Since labeled data only exists in the source domain, a model is expected to bridge the domain gap for tackling cross-domain ABSA. Though domain adaptation methods have proven to be effective, most of them are based on a discriminative model, which needs to be specifically designed for different ABSA tasks. To offer a more general solution, we propose a unified bidirectional generative framework to tackle various cross-domain ABSA tasks. Specifically, our framework trains a generative model in both text-to-label and label-to-text directions. The former transforms each task into a unified format to learn domain-agnostic features, and the latter generates natural sentences from noisy labels for data augmentation, with which a more accurate model can be trained. To investigate the effectiveness and generality of our framework, we conduct extensive experiments on four cross-domain ABSA tasks and present new state-of-the-art results on all tasks. Our data and code are publicly available at \url{https://github.com/DAMO-NLP-SG/BGCA}.

cs.CL↗

arXiv2023

An integral sliding-mode parallel control approach for general nonlinear systems via piecewise affine linear models

The fundamental problem of stabilizing a general nonaffine continuous-time nonlinear system is investigated via piecewise affine linear models (PALMs) in this article. A novel integral sliding-mode parallel control (ISMPC) approach is developed, where an uncertain piecewise affine system (PWA) is constructed to model a nonaffine continuous-time nonlinear system equivalently on a compact region containing the origin. A piecewise sliding-mode parallel controller is designed to globally stabilize the PALM and, consequently, to semiglobally stabilize the original nonlinear system. The proposed scheme enjoys three favorable features: (i) some restrictions on the system input channel are eliminated, thus the developed method is more relaxed compared with the published approaches; (ii) it is convenient to be used to deal with both matched and unmatched uncertainties of the system; and (iii) the proposed piecewise parallel controller generates smooth control signals even around the boundaries between different subspaces, which makes the developed control strategy more implementable and reliable. Moreover, we provide discussions about the universality analysis of the developed control strategy for two kinds of typical nonlinear systems. Simulation results from two numerical examples further demonstrate the performance of the developed control approach.

eess.SY↗

arXiv2023

Dif-Fusion: Towards High Color Fidelity in Infrared and Visible Image Fusion with Diffusion Models

Color plays an important role in human visual perception, reflecting the spectrum of objects. However, the existing infrared and visible image fusion methods rarely explore how to handle multi-spectral/channel data directly and achieve high color fidelity. This paper addresses the above issue by proposing a novel method with diffusion models, termed as Dif-Fusion, to generate the distribution of the multi-channel input data, which increases the ability of multi-source information aggregation and the fidelity of colors. In specific, instead of converting multi-channel images into single-channel data in existing fusion methods, we create the multi-channel data distribution with a denoising network in a latent space with forward and reverse diffusion process. Then, we use the the denoising network to extract the multi-channel diffusion features with both visible and infrared information. Finally, we feed the multi-channel diffusion features to the multi-channel fusion module to directly generate the three-channel fused image. To retain the texture and intensity information, we propose multi-channel gradient loss and intensity loss. Along with the current evaluation metrics for measuring texture and intensity fidelity, we introduce a new evaluation metric to quantify color fidelity. Extensive experiments indicate that our method is more effective than other state-of-the-art image fusion methods, especially in color fidelity.

cs.CV↗

arXiv2022

The Sabatier principle for Battery Anodes: Chemical Kinetics and Reversible Electrodeposition at Heterointerfaces

How surface chemistry influences reactions occurring thereupon has been a long-standing question of broad scientific and technological interest for centuries. Recently, it has re-emerged as a critical question in a subdiscipline of chemistry - electrochemistry at heterointerphases, where the answers have implications for both how, and in what forms, humanity stores the rising quantities of renewable electric power generated from solar and wind installations world-wide. Here we consider the relation between the surface chemistry at such interphases and the reversibility of electrochemical transformations at a rechargeable battery electrode. Conventional wisdom holds that stronger chemical interaction between the metal deposits and electrode promotes reversibility. We report instead that a moderate strength of chemical interaction between the deposit and the substrate, neither too weak nor too strong, enables highest reversibility and stability of the plating/stripping redox processes at a battery anode. Analogous to the empirical Sabatier principle for chemical heterogeneous catalysis, our finding arises from the confluence of competing processes - one driven by electrochemistry and the other by chemical alloying. Based on experimental evaluation of metal plating/stripping systems in battery anodes of contemporary interest, we show that such knowledge provides a powerful tool for designing key materials in highly reversible electrochemical energy storage technologies based on earth-abundant, low-cost metals.

physics.chem-ph↗

arXiv2022

Determinable and interpretable network representation for link prediction

As an intuitive description of complex physical, social, or brain systems, complex networks have fascinated scientists for decades. Recently, to abstract a network's structural and dynamical attributes for utilization, network representation has been one focus, mapping a network or its substructures (like nodes) into a low-dimensional vector space. Since the current methods are mostly based on machine learning, a black box of an input-output data fitting mechanism, generally the space's dimension is indeterminable and its elements are not interpreted. Although massive efforts to cope with this issue have included, for example, automated machine learning by computer scientists and computational theory by mathematics, the root causes still remain unresolved. Given that, from a physical perspective, this article proposes two determinable and interpretable node representation methods. To evaluate their effectiveness and generalization, this article further proposes Adaptive and Interpretable ProbS (AIProbS), a network-based model that can utilize node representations for link prediction. Experimental results showed that the AIProbS can reach state-of-the-art precision beyond baseline models, and by and large it can make a good trade-off with machine learning-based models on precision, determinacy, and interpretability, indicating that physical methods could also play a large role in the study of network representation.

cs.SI↗

arXiv2022

"Help! Can You Hear Me?": Understanding How Help-Seeking Posts are Overwhelmed on Social Media during a Natural Disaster

Posting help-seeking requests on social media has been broadly adopted by victims during natural disasters to look for urgent rescue and supplies. The help-seeking requests need to get sufficient public attention and be promptly routed to the intended target(s) for timely responses. However, the huge volume and diverse types of crisis-related posts on social media might limit help-seeking requests to receive adequate engagement and lead to their overwhelm. To understand this problem, this work proposes a mixed-methods approach to figure out the overwhelm situation of help-seeking requests, and individuals' and online communities' strategies to cope. We focused on the 2021 Henan Floods in China and collected 141,674 help-seeking posts with the keyword "Henan Rainstorm Mutual Aid" on a popular Chinese social media platform Weibo. The findings indicate that help-seeking posts confront critical challenges of both external overwhelm (i.e., an enormous number of non-help-seeking posts with the help-seeking-related keyword distracting public attention) and internal overwhelm (i.e., attention inequality with 5% help-seeking posts receiving more than 95% likes, comments, and shares). We discover linguistic and non-linguistic help-seeking strategies that could help to prevent the overwhelm, such as including contact information, disclosing situational vulnerabilities, using subjective narratives, and structuring help-seeking posts to a normalized syntax. We also illustrate how community members spontaneously work to prevent the overwhelm with their collective wisdom (e.g., norm development through discussion) and collaborative work (e.g., cross-community support). We reflect on how the findings enrich the literature in crisis informatics and raise design implications that facilitate effective help-seeking on social media during natural disasters.

cs.HC↗

arXiv2022

Recommender systems based on graph embedding techniques: A comprehensive review

As a pivotal tool to alleviate the information overload problem, recommender systems aim to predict user's preferred items from millions of candidates by analyzing observed user-item relations. As for alleviating the sparsity and cold start problems encountered by recommender systems, researchers resort to employing side information or knowledge in recommendation as a strategy for uncovering hidden (indirect) user-item relations, aiming to enrich observed information (or data) for recommendation. However, in the face of the high complexity and large scale of side information and knowledge, this strategy relies for efficient implementation on the scalability of recommendation models. Not until after the prevalence of machine learning did graph embedding techniques be a concentration, which can efficiently utilize complex and large-scale data. In light of that, equipping recommender systems with graph embedding techniques has been widely studied these years, appearing to outperform conventional recommendation implemented directly based on graph topological analysis. As the focus, this article retrospects graph embedding-based recommendation from embedding techniques for bipartite graphs, general graphs and knowledge graphs, and proposes a general design pipeline of that. In addition, after comparing several representative graph embedding-based recommendation models with the most common-used conventional recommendation models on simulations, this article manifests that the conventional models can overall outperform the graph embedding-based ones in predicting implicit user-item interactions, revealing the comparative weakness of graph embedding-based recommendation in these tasks. To foster future research, this article proposes suggestions on making a trade-off between graph embedding-based recommendation and conventional recommendation in different tasks, and puts forward open questions.

cs.IR↗

arXiv2022

Hyperparameter-free and Explainable Whole Graph Embedding

Graphs can be used to describe complex systems. Recently, whole graph embedding (graph representation learning) can compress a graph into a compact lower-dimension vector while preserving intrinsic properties, earning much attention. However, most graph embedding methods have problems such as tedious parameter tuning or poor explanation. This paper presents a simple and hyperparameter-free whole graph embedding method based on the DHC (Degree, H-index, and Coreness) theorem and Shannon Entropy (E), abbreviated as DHC-E. The DHC-E can provide a trade-off between simplicity and quality for supervised classification learning tasks involving molecular, social, and brain networks. Moreover, it performs well in lower-dimensional graph visualization. Overall, the DHC-E is simple, hyperparameter-free, and explainable for whole graph embedding with promising potential for exploring graph classification and lower-dimensional graph visualization.

cs.LG↗

arXiv2021

An Alignment-Agnostic Model for Chinese Text Error Correction

This paper investigates how to correct Chinese text errors with types of mistaken, missing and redundant characters, which is common for Chinese native speakers. Most existing models based on detect-correct framework can correct mistaken characters errors, but they cannot deal with missing or redundant characters. The reason is that lengths of sentences before and after correction are not the same, leading to the inconsistence between model inputs and outputs. Although the Seq2Seq-based or sequence tagging methods provide solutions to the problem and achieved relatively good results on English context, but they do not perform well in Chinese context according to our experimental results. In our work, we propose a novel detect-correct framework which is alignment-agnostic, meaning that it can handle both text aligned and non-aligned occasions, and it can also serve as a cold start model when there are no annotated data provided. Experimental results on three datasets demonstrate that our method is effective and achieves the best performance among existing published models.

cs.CL↗

arXiv2021

NVAE-GAN Based Approach for Unsupervised Time Series Anomaly Detection

In recent studies, Lots of work has been done to solve time series anomaly detection by applying Variational Auto-Encoders (VAEs). Time series anomaly detection is a very common but challenging task in many industries, which plays an important role in network monitoring, facility maintenance, information security, and so on. However, it is very difficult to detect anomalies in time series with high accuracy, due to noisy data collected from real world, and complicated abnormal patterns. From recent studies, we are inspired by Nouveau VAE (NVAE) and propose our anomaly detection model: Time series to Image VAE (T2IVAE), an unsupervised model based on NVAE for univariate series, transforming 1D time series to 2D image as input, and adopting the reconstruction error to detect anomalies. Besides, we also apply the Generative Adversarial Networks based techniques to T2IVAE training strategy, aiming to reduce the overfitting. We evaluate our model performance on three datasets, and compare it with other several popular models using F1 score. T2IVAE achieves 0.639 on Numenta Anomaly Benchmark, 0.651 on public dataset from NASA, and 0.504 on our dataset collected from real-world scenario, outperforms other comparison models.

cs.LG↗

arXiv2020

Multi-Target Deep Learning for Algal Detection and Classification

Water quality has a direct impact on industry, agriculture, and public health. Algae species are common indicators of water quality. It is because algal communities are sensitive to changes in their habitats, giving valuable knowledge on variations in water quality. However, water quality analysis requires professional inspection of algal detection and classification under microscopes, which is very time-consuming and tedious. In this paper, we propose a novel multi-target deep learning framework for algal detection and classification. Extensive experiments were carried out on a large-scale colored microscopic algal dataset. Experimental results demonstrate that the proposed method leads to the promising performance on algal detection, class identification and genus identification.

cs.CV↗

arXiv2019

Letter: Mathematical Foundation of Turbulence Generation-Symmetric to Asymmetric Liutex/Rortex

Vortex has been considered as the building block and muscle of turbulence for long time. A new physical quantity called Liutex (previously named Rortex) has been defined as the rigid rotation part of fluid motion. From DNS and experiment, forest of hairpin vortices has been found in flow transition and low Reynolds turbulence, but more and more one leg vortices appear in fully developed turbulence. This letter shows hairpin vortex is unstable. Hairpin vortex will weaken or lose one leg by the shear and vortex interaction. This conclusion is made by Liutex definition and mathematical analysis without any physical assumptions. The asymmetry of vortex is caused by the interaction of symmetric shear and symmetric Liutex due to the definition of Liutex, which considers the smaller element of a pair of vorticity elements as the rotational strength. For a 2-D fluid rotation, if a disturbance shear effects the larger element, the rotation strength will not be changed, but if the disturbance shear effects the smaller element, the rotation strength will be immediate changed due to the Liutex strength definition. For a rigid rotation, if both vorticity of the shear and Liutex have the same directions, e.g., clockwise, the Liutex strength will not be changed. If the vorticity of the shear and Liutex have different directions, e.g., one clockwise and one counterclockwise, the Liutex strength will be weakened. Then, the hairpin vortex could lose the symmetry and even deform to a one leg vortex. The one leg vortex cannot keep balanced, and the chaotic motion and flow fluctuation are doomed. This is considered as the mathematical foundation of turbulence formation. This theory has been checked and proved by our DNS results of boundary layer transition.

physics.flu-dyn↗

arXiv2018

Nitric Oxide in Climatological Global Energy Budget During 1982-2013

Over the past decades, temperature and density of the upper atmosphere show negative trends and decrease of the upper atmospheric temperature is attributed to the declining neutral density. Specifically, nitric oxide (NO) and carbon dioxide (CO2) govern thermospheric cooling at 5.3 and 15 micron, respectively. While a lot of efforts have focused on the CO2 effects on the long-term trends, relatively less attention has been paid to the impacts by NO, which responds to solar and geomagnetic activities dynamically. In this study, we investigate the role of NO in climatological global energy budget for the recent three solar cycles using the Global Ionosphere Thermosphere Model. From 1982 to 2013, the F10.7 and Ap indices showed a decadal decrease of ~8% and ~20%, respectively. By imposing temporal-varying F10.7 and Ap values in the simulations, we find a decadal change of -0.28x1011 W or -17.3% in total NO cooling power, which agrees well with that (-0.34x1011 W or -17.2%) from the empirical Thermosphere Climate Index derived from the TIMED/SABER data. Neutral density decreases by 10-20% at 200-450km and Tex decreases by 25.3 K per decade. The deduced-decadal change of NO cooling reaches ~25% of the sum of total heating at ~130 km and its significance decreases with altitude.

physics.space-ph↗

arXiv2018

Comparison of the Thermospheric Nitric Oxide Emission Observations and the Global Ionosphere-Thermosphere Model (GITM) Simulations: Sensitivity to Solar and Geomagnetic Activities

An accurate estimate of the energy budget (heating and cooling) of the ionosphere and thermosphere, especially during space weather events, has been a challenge. The abundance of Nitric Oxide (NO), a minor species in the thermosphere, is an important component of energy balance here because its production comes from energy sources able to break the strong bond of molecular nitrogen, and infrared emissions from NO play an important role in thermospheric cooling. Recent studies have significantly improved our understanding of NO chemistry and its relationship to energy deposition in the thermospheric photochemical reactions. In this study, the chemical scheme in the Global Ionosphere Thermosphere Model (GITM) is updated to better predict the lower thermospheric NO responses to solar and geomagnetic activity. We investigate the sensitivity of the 5.3-micron NO emission to F10.7 and Ap indices by comparing the global integrated emission from GITM with an empirical proxy derived from the Sounding of the Atmosphere using Broadband Emission Radiometry measurements. GITM's total emission agrees well within 20% of the empirical values. The updated chemistry scheme significantly elevates the level of integrated emission compared to the previous scheme. The inclusion of N2(A)-related production of NO contributes an additional 5-25% to the emission. Localized enhancement of ~70% in column density and a factor of three in column emission are simulated at a moderate geomagnetic level.

physics.space-ph↗

arXiv2018

An Algorithmic Information Calculus for Causal Discovery and Reprogramming Systems

We demonstrate that the algorithmic information content of a system is deeply connected to its potential dynamics, thus affording an avenue for moving systems in the information-theoretic space and controlling them in the phase space. To this end we performed experiments and validated the results on (1) a very large set of small graphs, (2) a number of larger networks with different topologies, and (3) biological networks from a widely studied and validated genetic network (e.coli) as well as on a significant number of differentiating (Th17) and differentiated human cells from high quality databases (Harvard's CellNet) with results conforming to experimentally validated biological data. Based on these results we introduce a conceptual framework, a model-based interventional calculus and a reprogrammability measure with which to steer, manipulate, and reconstruct the dynamics of non- linear dynamical systems from partial and disordered observations. The method consists in finding and applying a series of controlled interventions to a dynamical system to estimate how its algorithmic information content is affected when every one of its elements are perturbed. The approach represents an alternative to numerical simulation and statistical approaches for inferring causal mechanistic/generative models and finding first principles. We demonstrate the framework's capabilities by reconstructing the phase space of some discrete dynamical systems (cellular automata) as case study and reconstructing their generating rules. We thus advance tools for reprogramming artificial and living systems without full knowledge or access to the system's actual kinetic equations or probability distributions yielding a suite of universal and parameter-free algorithms of wide applicability ranging from causation, dimension reduction, feature selection and model generation.

q-bio.OT↗

arXiv2017

HiDi: An efficient reverse engineering schema for large scale dynamic regulatory network reconstruction using adaptive differentiation

The use of differential equations (ODE) is one of the most promising approaches to network inference. The success of ODE-based approaches has, however, been limited, due to the difficulty in estimating parameters and by their lack of scalability. Here we introduce a novel method and pipeline to reverse engineer gene regulatory networks from gene expression of time series and perturbation data based upon an improvement on the calculation scheme of the derivatives and a pre-filtration step to reduce the number of possible links. The method introduces a linear differential equation model with adaptive numerical differentiation that is scalable to extremely large regulatory networks. We demonstrate the ability of this method to outperform current state-of-the-art methods applied to experimental and synthetic data using test data from the DREAM4 and DREAM5 challenges. Our method displays greater accuracy and scalability. We benchmark the performance of the pipeline with respect to data set size and levels of noise. We show that the computation time is linear over various network sizes.

q-bio.MN↗