- Topic Modeling
- Natural Language Processing Techniques
- CO2 Reduction Techniques and Catalysts
- Digital and Cyber Forensics
- Speech Recognition and Synthesis
- AI in cancer detection
- Nanocluster Synthesis and Applications
- Algorithms and Data Compression
- Privacy-Preserving Technologies in Data
- Advanced Neural Network Applications
- Polyoxometalates: Synthesis and Applications
- Ionic liquids properties and applications
- Cryptography and Data Security
- Carbon dioxide utilization in catalysis
- Non-Destructive Testing Techniques
- Multimodal Machine Learning Applications
- Misinformation and Its Impacts
- Advanced MRI Techniques and Applications
- Lanthanide and Transition Metal Complexes
- Digital and Traditional Archives Management
- Cardiac Valve Diseases and Treatments
- Speech and dialogue systems
- Web Data Mining and Analysis
- Data Quality and Management
- Advanced Text Analysis Techniques
Northwest A&F University
2025
Xi'an Jiaotong University
2024
Robert Bosch (Germany)
2022-2024
Hainan University
2021-2024
Tencent (China)
2018-2023
Alibaba Group (China)
2020-2021
Weatherford College
2021
Ping An (China)
2020
China Railway 18th Bureau Group Corporation
2020
China Railway Group (China)
2020
Pathology Artificial Intelligence Platform (PAIP) is a free research platform in support of pathological artificial intelligence (AI). The main goal the to construct high-quality pathology learning data set that will allow greater accessibility. PAIP Liver Cancer Segmentation Challenge, organized conjunction with Medical Image Computing and Computer Assisted Intervention Society (MICCAI 2019), first image analysis challenge apply datasets. was evaluate new existing algorithms for automated...
Rumor spreaders are increasingly utilizing multimedia content to attract the attention and trust of news consumers. Though quite a few rumor detection models have exploited multi-modal data, they seldom consider inconsistent semantics between images texts, rarely spot inconsistency among post contents background knowledge. In addition, commonly assume completeness multiple modalities thus incapable handling handle missing in real-life scenarios. Motivated by intuition that rumors social...
Jianqiang Ma, Erhard Hinrichs. Proceedings of the 53rd Annual Meeting Association for Computational Linguistics and 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2015.
Natural small molecule compounds play crucial roles in regulating fat deposition. Beta-sitosterol exhibits multiple biological activities such as cholesterol reduction and anticancer effects. However, its regulatory mechanism the differentiation of bovine preadipocytes remains unclear. We identified potential associations with processes regulation lipid metabolism through prediction targets. utilized techniques Oil Red O staining, Western blotting, RNA-seq, others to elucidate promoting...
Applications of Internet Vehicles (IoV) make the life human beings more intelligent and convenient. However, in present, there are some problems IoV, such as data silos poor privacy preservation. To address challenges we propose a blockchain-based federated learning pool (BFLP) framework. BFLP allows models to be trained without sharing raw data, it can choose most suitable method according actual application scenarios. Considering computing power vehicle systems, construct lightweight...
Rumor spreaders are increasingly utilizing multimedia content to attract the attention and trust of news consumers. Though a set rumor detection models have exploited multi-modal data, they seldom consider inconsistent relationships among images texts. Moreover, also fail find powerful way spot inconsistency information post contents background knowledge. Motivated by intuition that rumors more likely in semantics, novel Knowledge-guided Dual-inconsistency network is proposed detect with...
We propose a simple and effective few-shot model for slot tagging.Recent work shows that it is promising to extend standard fewshot classification methods sequence labeling with CRF-specific augmentations.Such show strengths in encoding name semantics dependencies.However, we find these can be obtained by much simpler method, which casts tagging into machine reading comprehension (MRC).We fine-tune BERT-based MRC mixture of source domain (few-shot) target data.Such method outperforms...
Error mining is a useful technique for identifying forms that cause incomplete parses of sentences. We extend the iterative method Sagot and de la Clergerie (2006) to treat n-grams an arbitrary length. An inherent problem incorporating longer data sparseness. Our new takes sparseness into account, producing are as long necessary identify problematic forms, but not longer.
The Double Random Phase Encoding (DRPE) image encryption method has garnered significant attention in color processing and optical encryption, thanks to its parallel of R, G, B. However, DRPE-based faces two challenges. Firstly, it disregards the correlation B, compromising encrypted image's robustness. Secondly, DRPE schemes relying on Discrete Fourier Transform (DFT) Fractional (DFRFT) are vulnerable linear attacks, such as Known Plaintext Attack (KPA) Chosen (CPA). Quantum walk...
Abstract Metaheuristic is one of the techniques to improve security image encryption. However, existing metaheuristic cryptosystems based on may have convergence difficulties during optimization process, which cause insecurity and slow convergence. Besides, time cost parallel execution model applied not low enough. Therefore, a many‐objective optimized key generation framework proposed. Firstly, if four or more indicators cryptosystem, are results test, need be optimized, algorithm employed...
On the WikiSQL benchmark, state-of-the-art text-to-SQL systems typically take a slot- filling approach by building several dedicated models for each type of slots. Such modularized are not only complex but also limited capacity capturing inter-dependencies among SQL clauses. To solve these problems, this paper proposes novel extraction-linking approach, where unified extractor recognizes all types slot mentions appearing in question sentence before linker maps recognized columns to table...
For languages such as German where compounds occur frequently and are written single tokens, a wide variety of NLP applications benefits from recognizing splitting compounds.As the traditional word frequency-based approach to compound has several drawbacks, this paper introduces letter sequence labeling approach, which can utilize rich form features build discriminative learning models that optimized for splitting.Experiments show proposed method significantly outperforms state-ofthe-art splitters.
This paper presents a novel model that learns and exploits embeddings of phone ngrams for word segmentation in child language acquisition.Embedding-based models are evaluated on phonemically transcribed corpus child-directed speech, comparison with their symbolic counterparts using the common learning framework features.Results show significantly improves performance.We make use extensive visualization to understand what has learned.We learned informative both phonology general.
Prepostitional phrase (PP) attachment is a well known challenge to parsing. In this paper, we combine the insights of different works, namely: (1) treating PP as classification task with an arbitrary number candidates; (2) using auxiliary distributions augment data beyond hand-annotated training set; (3) topological fields get information about distribution throughout clauses and (4) state-of-the-art techniques such word embeddings neural networks. We show that jointly these leads...
Accurately segmenting left atrium in MR volume can benefit the ablation procedure of atrial fibrillation. Traditional automated solutions often fail relieving experts from labor-intensive manual labeling. In this paper, we propose a deep neural network based solution for segmentation gadolinium-enhanced volumes with promising performance. We firstly argue that, volumetric task, networks 2D fashion present great superiorities time efficiency and accuracy than 3D fashion. Considering highly...
Text-to-SQL systems offers natural language interfaces to databases, which can automatically generates SQL queries given questions. On the WikiSQL benchmark, state-of- the-art text-to-SQL typically take a slot-filling approach by building several specialized models for each type of slot. Despite being effective, such modularized are complex and also fall short in jointly learning different slots. To solve these problems, this paper proposes novel that formulates task as question answering...
The issue of trust in the management digital records has been a topic research for number years. During this time most researchers have concentrated on nature and meaning record itself, rather than potential use as evidence originators' origins, functions, activities. Through comparison real world world, we demonstrate importance evidential value assessment trustworthiness record. In paper investigate, identify specify requirements value, based our life cycle model Finally, show briefly how...
Unique stimuli in organelles could trigger situ self-assembly of nanofibers with ordered structures, thus exhibiting differential influences on cells.In this minireview, we feature organelles, highlight their formation mechanism and triggering strategies, discuss the biomedical applications current challenges.Nucleic acid nanostructures are capable precisely arranging proteins at nanoscale distances have proven to be powerful scaffolds for studying mechanistic effects enzyme cascades.This...
Angewandte Chemie International Edition is a journal of the Gesellschaft Deutscher Chemiker (GDCh), largest chemistryrelated scientific society in continental Europe.Information on various activities and services GDCh, for example, cheaper subscription to Edition, as well applications membership can be found at www.gdch.deor requested from Postfach 900440, D-60444 Frankfurt am Main, Germany.… 2D-silica nanosheets results formation micron-sized soccer ball-like hollow shells, which...
In this paper, we introduce Hunyuan-Large, which is currently the largest open-source Transformer-based mixture of experts model, with a total 389 billion parameters and 52 activation parameters, capable handling up to 256K tokens. We conduct thorough evaluation Hunyuan-Large's superior performance across various benchmarks including language understanding generation, logical reasoning, mathematical problem-solving, coding, long-context, aggregated tasks, where it outperforms LLama3.1-70B...