A5100714256- Topic Modeling
- Advanced Image and Video Retrieval Techniques
- Data Mining Algorithms and Applications
- Market Dynamics and Volatility
- Data Management and Algorithms
- Energy Load and Power Forecasting
- Computer Graphics and Visualization Techniques
- Neural Networks and Applications
- Rough Sets and Fuzzy Logic
- Remote-Sensing Image Classification
- Stock Market Forecasting Methods
- Advanced Clustering Algorithms Research
- Natural Language Processing Techniques
- Computational Drug Discovery Methods
- 3D Shape Modeling and Analysis
- Fault Detection and Control Systems
- Image Processing and 3D Reconstruction
- Pharmaceutical Quality and Counterfeiting
- Machine Learning in Bioinformatics
- Multimodal Machine Learning Applications
- Automated Road and Building Extraction
- Image Processing Techniques and Applications
- Advanced Image Processing Techniques
- Image and Signal Denoising Methods
- Innovative Microfluidic and Catalytic Techniques Innovation
Nanjing University of Posts and Telecommunications
2022-2023
Shanghai Normal University
2023
Tomorrows Children’s Fund
2021
Nanjing University of Aeronautics and Astronautics
2017
Tongji University
2015-2017
Tiangong University
2014
Zhejiang University
2012-2013
III-N Technology (United States)
2012
Jiangsu University
2010
Shanghai University of Finance and Economics
2010
Heart disease is one of the most common diseases in world. The objective this study to aid diagnosis heart using a hybrid classification system based on ReliefF and Rough Set (RFRS) method. proposed contains two subsystems: RFRS feature selection with an ensemble classifier. first includes three stages: (i) data discretization, (ii) extraction algorithm, (iii) reduction heuristic algorithm that we developed. In second system, classifier C4.5 Statlog (Heart) dataset, obtained from UCI...
Unsupervised disentanglement learning is a crucial issue for understanding and exploiting deep generative models. Recently, SeFa tries to find latent disentangled directions by performing SVD on the first projection of pretrained GAN. However, it only applied layer works in post-processing way. Hessian Penalty minimizes off-diagonal entries output's matrix facilitate disentanglement, can be multi-layers. constrains each entry output independently, making not sufficient disentangling (e.g.,...
This report provides an overview of the challenge hosted at OpenSUN3D Workshop on Open-Vocabulary 3D Scene Understanding held in conjunction with ICCV 2023. The goal this workshop series is to provide a platform for exploration and discussion open-vocabulary scene understanding tasks, including but not limited segmentation, detection mapping. We workshop, present dataset, evaluation methodology, brief descriptions winning methods. For additional details, please see...
For most of data sets, there exist some redundant, irrelevant and even noise features. Usually, are plenty features in medical sets the correlation among is strong. So, feature selection gets great concern recent years. RELIEFF one effective algorithms, but cannot remove redundant RS a mathematical approach to intelligent analysis can novel RS- algorithm proposed this paper. In RS-RELIEFF, reduction applied set with firstly, then later, new integrative weight each condition will be got end....
The unstructured road detection plays a key role in an autonomous vehicle navigation system. However, the images often contain shadows and are easily affected by ambient light, resulting to inaccuracy with detection. A robust technique is required. In this paper, we adopted improved fuzzy c-means(FCM) clustering algorithm address these issues. new considered neighborhood impact factor when calculating distances between cluster center pixel. Our experimental results show that FCM have better outcomes.
Most of the existing compositional generalization datasets are synthetically-generated, resulting in a lack natural language variation. While there have been recent attempts to introduce non-synthetic for generalization, they suffer from either limited data scale or diversity forms combinations. To better investigate with more linguistic phenomena and diversity, we propose DIsh NamE Recognition (DiNeR) task create large realistic Chinese dataset. Given recipe instruction, models required...
Although Large Language Models (LLMs) are becoming increasingly powerful, they still exhibit significant but subtle weaknesses, such as mistakes in instruction-following or coding tasks. As these unexpected errors could lead to severe consequences practical deployments, it is crucial investigate the limitations within LLMs systematically. Traditional benchmarking approaches cannot thoroughly pinpoint specific model deficiencies, while manual inspections costly and not scalable. In this...
The long short‐term memory (LSTM) recurrent neural network algorithm in deep learning has demonstrated significant superiority predicting the realized volatility (RV) of crude oil prices. However, there is no robust and consistent conclusion regarding handling microstructural noise from high‐frequency data during prediction process. Therefore, this study utilizes six commonly used decomposition methods, as documented literature, to address issue decompose RV series Chinese futures....
Large Language Models (LLMs) have demonstrated notable capabilities across various tasks, showcasing complex problem-solving abilities. Understanding and executing rules, along with multi-step planning, are fundamental to logical reasoning critical for practical LLM agents decision-making systems. However, evaluating LLMs as effective rule-based executors planners remains underexplored. In this paper, we introduce LogicGame, a novel benchmark designed evaluate the comprehensive rule...
The Original TOPSIS (Technique for Order Preference by Similarity to an Ideal Solution) Method is a ranking method based on the principle that chosen alternative should be as close ideal solution, rather than negative-ideal possible. However, currently utilizing in establishing evaluation system of indicator-based model, failed take weights indices into account. Besides, other weighting like Subjective Evaluation proved largely influenced various subjective situations. To resolve problem...
The author attempts to identify how the image of state is formed on example China. article reveals theoretical approaches concept “the state” and demonstrates it can be used in practice. In particular, examines issues symbols, stereotypes that are by state. analyzes influence culture as most important tool “soft power” notes significance semantic content political image, which compared with a kind source, at subconscious level sets certain angle perception new information about this object....
The generation of bas-relief based on 3D models has always been a research hotspot in computer graphics. Traditional algorithms are computationally complex. latest methods deep learning bring new ideas to the problem but rely supervised training, which requires large number samples be constructed manually. In this paper, an unsupervised method is proposed. method, we take depth field point cloud as input, and reconstruct using Convolutional Neural Networks. During process, smallest change...