A5011482849- Music and Audio Processing
- Speech and Audio Processing
- Music Technology and Sound Studies
- Multimodal Machine Learning Applications
- Topic Modeling
- Natural Language Processing Techniques
- Speech Recognition and Synthesis
- Recommender Systems and Techniques
- Plant Virus Research Studies
- Retinal Imaging and Analysis
- Retinal and Optic Conditions
- Asymmetric Hydrogenation and Catalysis
- Chemical synthesis and alkaloids
- Alkaloids: synthesis and pharmacology
- Asymmetric Synthesis and Catalysis
- Transgenic Plants and Applications
- Marine Sponges and Natural Products
- Optical Coherence Tomography Applications
- Advanced Computational Techniques and Applications
- Authorship Attribution and Profiling
- Plant and Fungal Interactions Research
- Plant Disease Resistance and Genetics
- Sex work and related issues
- Fire Detection and Safety Systems
- Explainable Artificial Intelligence (XAI)
Donghua University
2022-2025
Jinan University
2020-2024
Northern (Arctic) Federal University
2022
Fudan University
2000-2022
Xinjiang University
2021
Shaanxi Normal University
2020
Shenzhen Institutes of Advanced Technology
2018-2019
Chinese Academy of Sciences
2018
Northeast Agricultural University
2018
Seoul National University
2011-2017
OPINION article Front. Digit. Health, 26 June 2020 | https://doi.org/10.3389/fdgth.2020.00005
Musical audio is generally composed of three physical properties: frequency, time and magnitude. Interestingly, human auditory periphery also provides neural codes for each these dimensions to perceive music. Inspired by intrinsic characteristics, a frequency-temporal attention network proposed mimic singing melody extraction. In particular, the model contains modules selective fusion module corresponding properties. The frequency used select same activation bands as did in cochlear temporal...
The structure of wondonin marine natural products was renovated to attain new drug-like scaffolds. Wondonins have novel antiangiogenic properties without overt cytotoxicity. However, the chemical instability and synthetic complexity wondonins hindered their development as a type antiangiogenesis agent. Using structure-based bioisosterism, benzodioxole moiety changed benzothiazole, imidazole replaced by 1,2,3-triazole. Our efforts resulted in scaffold with enhanced activity minimized One...
Singing melody extraction is an important task in the field of music information retrieval (MIR). The development data-driven models for this have achieved great successes. However, existing two major limitations: firstly, most singing formulated as a pixel-level prediction task. lack labeling data has limited model further improvements. Secondly, generalization are prone to be disturbed by genres. To address issues mentioned above, paper, we propose multi-Task contrastive learning framework...
In this paper, we develop a neural attentive interpretable recommendation system, named NAIRS. A self-attention network, as key component of the is designed to assign attention weights interacted items user. This mechanism can distinguish importance various in contributing user profile. %, and it also provides recommendations. Based on profiles obtained by NAIRS offers personalized high-quality recommendation. Moreover, develops visual cues interpret demo application with implementation...
Attention-based recurrent neural network models for joint intent detection and slot filling have achieved a state-of-the-art performance. Most previous works exploited semantic level information to calculate the attention weights. However, few taken importance of word into consideration. In this paper, we propose WAIS, filling. Considering that strong relationship, further fusion gate integrates together jointly training two tasks. Extensive experiments show proposed model has robust...
Singing melody extraction from polyphonic music is a critical and challenging task in information retrieval (MIR). However, due to the interfere of accompaniment background noise, it key obtain global semantic representation that discriminates singing line. To address this issue, we consider two aspects regards obtaining representation: relationships spectrum between channels. In paper, propose novel hierarchical graph-based network for extraction. particular, according its characteristics...
A plethora of great successes has been achieved by the existing convolutional neural networks (CNN) for respiratory sound classification. Nevertheless, simultaneously capturing both local and global features can never be an easy task due to limitation a CNN's structure. In this contribution, we propose novel glance-and-gaze network address aforementioned issue. The glance block aims learn information, while gaze is responsible learning patterns suppressing noises that attenuates final...
The first total syntheses of (−)-isowondonin A and B, which are unusual imidazole marine alkaloids, has been accomplished through the development methods for selective formation styryl sulfate group regioselective alkylation imidazole. Application Noyori asymmetric hydrogenation ketones allows synthesis. These results in conjunction with ECD calculations led to determination absolute configuration isowondonins.
Singing melody extraction in polyphonic musical audio is a very critical and challenging task music information retrieval (MIR). Contextual frame-level has proven its effectiveness this task. However, existing works assign equal weight to each contextual frame, which may hinder the further improvement of performance. To end, we propose hierarchical attention network for singing (HANME) extract discriminative attention-aware features alleviate workload convolutional recurrent neural (CRNN)...
The mammalian target of rapamycin (mTOR) is a critical signaling hub for sustaining cancer survival. Targeting mTOR and inducing autophagic cell death downstream it represent promising therapeutic strategies prevention. A US Food Drug Administration-approved drug library containing 616 small molecules used to screen anticancer drugs against colorectal (CRC) cells that rely on mTOR. This led the identification an antipsychotic aripiprazole, which significantly induced inhibition apoptosis in...
Intent detection and slot filling are two important tasks in Spoken Language Understanding. The Condition Random Fields (CRF) was introduced for the pretty much same fashion to deep neural networks. Recently, attention based encoder-decoder models have shown promising results joint intent spoken language understanding dialog systems. However, often trained separately. In this paper, we propose ACJIS, a novel Attentive Cross approach Joint Slot filling. We introduce cross enhance modeling...
A method for the stereoselective formation of enol sulfate esters from α-aryl aldehydes is described. This involved stereocontrolled enolization a carbonyl group with DBU or t-BuOK followed by trapping reactive sulfuryl imidazolium salt, providing corresponding styryl in good to excellent yields and stereoselectivities. was applied first total synthesis natural products stereocomplementary manner.
Author profiling is an important but challenging task. In this paper, we propose a novel Multi-Task learning framework for Profiling (MTAP), in which document modeling module shared across three different author tasks (i.e., age, gender and job classification tasks). To further boost profiling, integrate hierarchical features learned by models. Concretely, employ CNN, LSTM topic model to learn the character-level, word-level topic-level features, respectively. MTAP thus leverages benefits of...
By combining artificial neural network with deep learning technology, convolution is characterized by local perception, adaptive feature extraction and end-to-end application, etc., it has been used in image recognition target detection more recent years. Problems are existing widely the traditional safety helmet algorithm generally such as severe background interference, complex computing, high time-complexity largely fluctuant accuracy. A detective method for based on was proposed this...