A5100380647- Domain Adaptation and Few-Shot Learning
- Advanced Computational Techniques and Applications
- Service-Oriented Architecture and Web Services
- AI in cancer detection
- Topic Modeling
- Machine Learning in Healthcare
- Privacy-Preserving Technologies in Data
- Recommender Systems and Techniques
- Image and Signal Denoising Methods
- Data Stream Mining Techniques
- Advanced Electrical Measurement Techniques
- Generative Adversarial Networks and Image Synthesis
- Remote Sensing and LiDAR Applications
- COVID-19 diagnosis using AI
- Advanced Neural Network Applications
- Natural Language Processing Techniques
- Structural Health Monitoring Techniques
- Advanced Data Processing Techniques
- Pickering emulsions and particle stabilization
- Blasting Impact and Analysis
- Semantic Web and Ontologies
- Advanced Clustering Algorithms Research
- Music and Audio Processing
- Mathematics, Computing, and Information Processing
- Industrial Technology and Control Systems
China University of Geosciences (Beijing)
2024
Wuhan University of Science and Technology
2024
Ningxia University
2024
Shandong Normal University
2023
Liaoning University
2023
Yanching Institute of Technology
2023
Cavite State University
2023
Xi'an University of Architecture and Technology
2022
People’s Hospital of Rizhao
2022
Shandong Institute of Automation
2021
Abstract Gold precipitation in hydrothermal systems is traditionally attributed to supersaturation of gold due decreasing complex stability triggered by changes physicochemical conditions the ore fluid. However, ultrahigh-grade veins orogenic (shear zone related) deposits can contain kilograms per tonne or more, marked contrast typically very low concentrations (tens parts billion) The mineral assemblage commonly restricted native and/or Au-(Ag)-tellurides and occurs micro-fractures sheared...
Large language models (LLMs) have shown the potential to be integrated into human daily lives. Therefore, user preference is most critical criterion for assessing LLMs' performance in real-world scenarios. However, existing benchmarks mainly focus on measuring models' accuracy using multi-choice questions, which limits understanding of their capabilities real applications. We fill this gap by proposing a comprehensive Chinese benchmark SuperCLUE, named after another popular LLM CLUE....
Recent advancements in large language models (LLMs) have transformed the field of question answering (QA). However, evaluating LLMs medical is challenging due to lack standardized and comprehensive datasets. To address this gap, we introduce CMExam, sourced from Chinese National Medical Licensing Examination. CMExam consists 60K+ multiple-choice questions for objective evaluations, as well solution explanations model reasoning evaluation an open-ended manner. For in-depth analyses LLMs,...
Accurate estimation of customer lifetime value (LTV), which reflects the potential consumption a user over period time, is crucial for revenue management online advertising platforms. However, predicting LTV in real-world applications not an easy task since data usually insufficient within specific domain. To tackle this problem, we propose novel cross-domain adaptative framework (CDAF) to leverage from different domains. The proposed method able simultaneously mitigate scarce problem and...
We introduce SuperCLUE-Math6(SC-Math6), a new benchmark dataset to evaluate the mathematical reasoning abilities of Chinese language models. SC-Math6 is designed as an upgraded version GSM8K with enhanced difficulty, diversity, and application scope. It consists over 2000 word problems requiring multi-step providing natural solutions. propose innovative scheme quantify capability large models based on performance different steps. Experiments 13 representative demonstrate clear stratification...
Monaural source separation is an important research area which can help to improve the performance of several real-world applications, such as speech recognition and assisted living systems. Huang et al. proposed deep recurrent neural networks (DRNNs) with discriminative criterion objective function separation. However, penalty factor in selected randomly empirically. Therefore, we introduce approach calculate parameter term adaptively via discrepancy between target features. The be changed...
Over the past decade, domain adaptation has become a widely studied branch of transfer learning that aims to improve performance on target domains by leveraging knowledge from source domain. Conventional methods often assume access both and data simultaneously, which may not be feasible in real-world scenarios due privacy confidentiality concerns. As result, research Source-Free Domain Adaptation (SFDA) drawn growing attention recent years, only utilizes source-trained model unlabeled adapt...
Federated learning (FL) facilitates collaborative among multiple clients in a distributed manner, while ensuring privacy protection. However, its performance is inevitably degraded as suffering data heterogeneity, i.e., non-IID data. In this paper, we focus on the feature distribution skewed FL scenario, which widespread real-world applications. The main challenge lies shift caused by different underlying distributions of local datasets. While previous attempts achieved progress, few studies...
Federated Learning (FL) enables multiple clients to collaboratively learn in a distributed way, allowing for privacy protection. However, the real-world non-IID data will lead client drift which degrades performance of FL. Interestingly, we find that difference logits between local and global models increases as model is continuously updated, thus seriously deteriorating FL performance. This mainly due catastrophic forgetting caused by heterogeneity clients. To alleviate this problem,...
Semantic segmentation aims to divide a scene into regions with different semantic categories. The prevalent technique in is denoted as pixel-based segmentation, whereby classifications are assigned singular points. However, the principles used make predictions through these methods exhibit significant differences from way which would be processed human vision. When encountering new scenes, humans initially concentrate on each instance within three-dimensional scene, rather than individual...
Instance-incremental learning (IIL) focuses on continually with data of the same classes. Compared to class-incremental (CIL), IIL is seldom explored because suffers less from catastrophic forgetting (CF). However, besides retaining knowledge, in real-world deployment scenarios where class space always predefined, continual and cost-effective model promotion potential unavailability previous a more essential demand. Therefore, we first define new practical setting as promoting model's...
Based on the Fast Fourier Transform(FFT) of sine signal sampling sequence, this article proposed a self - adaptive frequency estimation method using multi spectral line information which is near peak map. To solve interpolation direction error rectangular windows Rife ratio method, combining with triangular new self-adaptively selects different methods according to frequency. Experimental simulation results show that has high accuracy and small calculation amount, less affected by noise...
JND threshold is usually used to control the embedding location and intensity in image watermarking system, determine extraction blind algorithm as well. A novel computation method based on low frequency coefficients of wavelet domain texture measures present enhance robustness threshold. Most embedded watermarks can be effectively located even when largely altered, thus it improve efficaciously detection result extraction. Besides, a called matching statistical characteristics given....
We investigate the linear complementarity problem with uncertain parameters (ULCP) which affect mapping affinely or quadratically. Assuming that distribution of belongs to some ambiguity set prescribed partial information, we formulate ULCP as a distributionally robust optimization reformulation named (DRCP), minimizes worst case an expected measure joint chance constraint probability being nonnegative is not less than given level. Applying cone dual theory and S-procedure, conservatively...
Based on the linear prediction (LP) property and high lags autocorrelation of sinusoidal signals, a new frequency estimators real sinusoid signal in additive white Gaussian noise is proposed. The estimator inspired by alternative derivation Pisarenko harmonic decomposer (PHD). A simple cubic function coefficients then employed for estimation. Computer simulations are included to show performance proposed via comparison with Cramer-Rao lower bound (CRLB) several conventional estimators.
Diffusion Probabilistic Models have recently shown remarkable performance in generative image modeling, attracting significant attention the computer vision community. However, while a substantial amount of diffusion-based research has focused on tasks, few studies applied diffusion models to general medical classification. In this paper, we propose first model (named DiffMIC) address classification by eliminating unexpected noise and perturbations images robustly capturing semantic...
College students’ management informatization ecosystem can help college students better manage information and resources, thus improving learning life efficiency. It’s important to introduce data technology into the student ecosystem, which realize extraction of behavior fluctuation achieve more targeted decisions. The current research lacks consideration its future development trend, such as impact application technologies, including artificial intelligence, big Fenix, Internet Things. This...