A5100340313- Natural Language Processing Techniques
- Topic Modeling
- Multimodal Machine Learning Applications
- Text Readability and Simplification
- Recommender Systems and Techniques
- Advanced Text Analysis Techniques
- Speech and dialogue systems
- Translation Studies and Practices
- Advanced Graph Neural Networks
- Lexicography and Language Studies
- Speech Recognition and Synthesis
- AI in cancer detection
- Sentiment Analysis and Opinion Mining
- Educational Technology and Pedagogy
- Epigenetics and DNA Methylation
- Text and Document Classification Technologies
- Algorithms and Data Compression
- Ideological and Political Education
- COVID-19 diagnosis using AI
- Hate Speech and Cyberbullying Detection
- Foreign Language Teaching Methods
- Complex Network Analysis Techniques
- Genomics and Rare Diseases
- Stock Market Forecasting Methods
- Video Analysis and Summarization
Shanghai Ninth People's Hospital
2024-2025
Shanghai Jiao Tong University
2013-2025
Auckland University of Technology
2022-2024
Chongqing University of Posts and Telecommunications
2023-2024
Shanghai University of Electric Power
2024
Alibaba Group (United States)
2023
Guangdong University of Finance
2023
Renmin University of China
2023
Beihang University
2016-2022
East China Jiaotong University
2022
In this paper, we describe a source-side reordering method based on syntactic chunks for phrase-based statistical machine translation. First, shallow parse the source language sentences. Then, rules are automatically learned from and word alignments. During translation, used to generate lattice each sentence. Experimental results reported Chinese-to-English task, showing an improvement of 0.5%--1.8% BLEU score absolute various test sets better computational efficiency than during decoding....
Abstract The precise mechanisms behind early embryonic arrest due to sperm‐related factors and the most effective strategies are not yet fully understood. Here, we present two cases of male infertility linked novel TDRD6 variants, associated with oligoasthenoteratozoospermia (OAT) arrest. To investigate underlying promising therapeutic approaches, Tdrd6 knock‐in knock‐out mice were generated. variant demonstrated OAT arrest, mirroring clinical observations our patients. Sperm from both...
Aligning large language models (LLMs) with human values, particularly in the face of stealthy and complex jailbreak attacks, presents a formidable challenge. In this study, we present simple yet highly effective defense strategy, i.e., Intention Analysis ($\mathbb{IA}$). The principle behind is to trigger LLMs' inherent self-correct improve ability through two-stage process: 1) essential intention analysis, 2) policy-aligned response. Notably, $\mathbb{IA}$ an inference-only method, thus...
This paper introduces new definitions of Chinese base phrases and presents a hybrid model to combine Memory-Based Learning method disambiguation proposal based on lexical information grammar rules populated from large corpus for 9 types chunking. Our experiment achieves an accuracy (F-measure) 93.4%. The significance the research lies in fact that it provides solid foundation parser.
In this paper, we present the KIT systems participating in Shared Translation Task translating between English↔German and English↔French.All translations are generated using phrase-based translation systems, different kinds of word-based, part-ofspeech-based cluster-based language models trained on provided data.Additional include bilingual models, reordering based part-of-speech tags syntactic parse trees, as well a lexicalized model.In order to make use noisy web-crawled data, apply...
With the in-depth development of China's education reform, combination educational public opinion and emotion analysis is conducive to discovery evolution law sentiment related factors, provides ideas for guidance relevant government departments. This paper proposes an model on policy network based text mining, obtains comment contents through Octopus crawler, preprocesses data, introduces calculation formula heat, which process divided into incubation period, outbreak spreading period...
Improving neural machine translation (NMT) systems with prompting has achieved significant progress in recent years. In this work, we focus on how to integrate multi-knowledge, multiple types of knowledge, into NMT models enhance the performance prompting. We propose a unified framework, which can effectively knowledge including sentences, terminologies/phrases and templates models. utilize as prefix-prompts input for encoder decoder guide process. The approach requires no changes model...
Abstract Neural Machine Translation (NMT) has achieved great developments in recent years, but we still have to face two challenges: establishing a high-quality corpus and exploring optimal parameters of models for long text translation. In this paper, first attempt set up paragraph-parallel based on English Chinese versions the novels then design hierarchical attention model it handle these challenges. Our encoder decoder take segmented clauses as input process words, clauses, paragraphs at...
In this paper, the KIT systems submitted to Shared Translation Task are presented.We participated in two translation directions: from German English and German.Both translations generated using phrase-based systems.The performance of was boosted by language models built based on different tokens such as word, part-of-speech, automacally word clusters.The difference order between is addressed part-of-speech syntactic tree-based reordering models.In addition a discriminative lexicon, we used...
Background: The histone deacetylase family of proteins, which includes the sirtuins, participates in a wide range cellular processes, and is intimately involved neurodegenerative illnesses. research on sirtuins has garnered lot interest. However, there are currently no effective therapeutic drugs. Methods: In order to explore potential inhibitors SIRTs, we first screened four lead compounds SIRT2 Traditional Chinese Medicine (TCM) for nervous disease using Auto- Dock Vina method. Then, with...
<title>Abstract</title> The prediction of stock market price trends has always been a challenging issue, attracting widespread attention from both economists and computer scientists. Recently, integrating prices with news data shown to be an effective strategy for enhancing the accuracy task. Yet, many current methods fail fully leverage intricate inter-stock relationships inherent in news.Applying deep learning, especially Graph Convolutional Networks (GCNs), predict demonstrated advanced...
In many challenging breast cancer pathology images, the proportion of truly informative tumor regions is extremely limited. The disparity between essential information required for clinical diagnosis (Tumor area less than 10) and vast amount data within Whole Slide Images (WSIs) makes it exceedingly difficult pathologists to identify subtle lesions. To address labor-intensive task imposed by this gap, paper proposes a dynamic sparse token based multi-instance learning framework. This...