Showing 1–7 of 7 results for author: Che, W

Improving Domain Generalization for Sound Classification with Sparse Frequency-Regularized Transformer

Authors: Honglin Mu, Wentian Xia, Wanxiang Che

Abstract: Sound classification models' performance suffers from generalizing on out-of-distribution (OOD) data. Numerous methods have been proposed to help the model generalize. However, most either introduce inference overheads or focus on long-lasting CNN-variants, while Transformers has been proven to outperform CNNs on numerous natural language processing and computer vision tasks. We propose FRITO, an… ▽ More Sound classification models' performance suffers from generalizing on out-of-distribution (OOD) data. Numerous methods have been proposed to help the model generalize. However, most either introduce inference overheads or focus on long-lasting CNN-variants, while Transformers has been proven to outperform CNNs on numerous natural language processing and computer vision tasks. We propose FRITO, an effective regularization technique on Transformer's self-attention, to improve the model's generalization ability by limiting each sequence position's attention receptive field along the frequency dimension on the spectrogram. Experiments show that our method helps Transformer models achieve SOTA generalization performance on TAU 2020 and Nsynth datasets while saving 20% inference time. △ Less

Submitted 18 July, 2023; originally announced July 2023.

Comments: Accepted by ICME 2023

Field Evaluation of Four Low-cost PM Sensors and Design, Development and Field Evaluation of A Wearable PM Exposure Monitoring System

Authors: Wei-Ying Yi, Yu Zhou, Ya-Fen Chan, Yee Leung, Kam-Sang Woo, Wen-Wei Che, Kai-Hon Lau, Jia-Min Chen, Kwong-Sak Leung

Abstract: To mitigate the significant biases/errors in research studying the associations between PM and health, which are introduced by the coarse/inadequate assessments of PM exposure from conventional PM monitoring paradigm, a personalized monitoring system consisting of a low-cost wearable PM device is proposed. However, due to the absence of a unifying evaluation protocol for low-cost PM sensors, the e… ▽ More To mitigate the significant biases/errors in research studying the associations between PM and health, which are introduced by the coarse/inadequate assessments of PM exposure from conventional PM monitoring paradigm, a personalized monitoring system consisting of a low-cost wearable PM device is proposed. However, due to the absence of a unifying evaluation protocol for low-cost PM sensors, the evaluation results/performance specifications from existing studies/datasheets are of limited reference values when attempting to determine the best candidate for the proposed system. In this regard, the authors appeal to the research community to develop a standardized evaluation protocol for low-cost PM sensors/devices, and a unifying attempt is established in this manuscript by adopting the definitive terminology from international documents and the evaluation metrics regarded as best practices. Collocated on the rooftop of the HKUST Supersite, four empirically selected PM sensors were compared against each other and calibrated against two reference monitors. They were then evaluated against the reference following the protocol. The PlanTower PMS-A003 sensor was selected for the wearable device as it outperformed the others in terms of affordability, portability, detection capability, data quality, as well as humidity and condensation insusceptibility. An automated approach was proposed to identify and remove the condensation associated abnormal measurements. The proposed device has better affordability and portability as well as similar usability and data accessibility compared to those existing devices recognized. The first 10 devices were also evaluated and calibrated at the Supersite. Additional 120 units were manufactured and delivered to the subjects to acquire their daily PM2.5 exposures for investigating the association with subclinical atherosclerosis. △ Less

Submitted 11 July, 2022; originally announced July 2022.

Analog cross coupled controller for oscillations: modeling and design via dominant system theory

Authors: Weiming Che, Thomas Chaffey, Fulvio Forni

Abstract: We propose a new analog feedback controller based on the classical cross coupled electronic oscillator. The goal is to drive a linear passive plant into oscillations. We model the circuit as Lur'e system and we derive a new graphical condition to certify oscillations (inverse circle criterion for dominance theory). These conditions are then specialized to minimal control architectures like RLC and… ▽ More We propose a new analog feedback controller based on the classical cross coupled electronic oscillator. The goal is to drive a linear passive plant into oscillations. We model the circuit as Lur'e system and we derive a new graphical condition to certify oscillations (inverse circle criterion for dominance theory). These conditions are then specialized to minimal control architectures like RLC and RC networks, and are illustrated with an example based on a DC motor model. △ Less

Submitted 30 March, 2022; originally announced March 2022.

Dominant Mixed Feedback Design for Stable Oscillations

Authors: Weiming Che, Fulvio Forni

Abstract: We present a design framework to induce stable oscillations through mixed feedback control. We provide conditions on the feedback gain and on the balance between positive and negative feedback contributions to guarantee robust oscillations. Using linear matrix inequalities, we later derive a systematic design for robustness to bounded dynamic uncertainties and for passive interconnections. The res… ▽ More We present a design framework to induce stable oscillations through mixed feedback control. We provide conditions on the feedback gain and on the balance between positive and negative feedback contributions to guarantee robust oscillations. Using linear matrix inequalities, we later derive a systematic design for robustness to bounded dynamic uncertainties and for passive interconnections. The results of the paper provide a system-theoretic justification to several observations from system biology and neuroscience pointing at the mixed feedback as a fundamental enabler for robust oscillations. Our results are illustrated through a distributed electrical circuit mimicking (simplified) neural dynamics. △ Less

Submitted 8 May, 2023; v1 submitted 13 October, 2021; originally announced October 2021.

Shaping oscillations via mixed feedback

Authors: W. Che, F. Forni

Abstract: We study the problem of controlling oscillations in closed loop by combining positive and negative feedback in a mixed configuration. We develop a complete design procedure to set the relative strength of the two feedback loops to achieve steady oscillations. The proposed design takes advantage of dominance theory and adopts classical harmonic balance and fast/slow analysis to regulate the frequen… ▽ More We study the problem of controlling oscillations in closed loop by combining positive and negative feedback in a mixed configuration. We develop a complete design procedure to set the relative strength of the two feedback loops to achieve steady oscillations. The proposed design takes advantage of dominance theory and adopts classical harmonic balance and fast/slow analysis to regulate the frequency of oscillations. The design is illustrated on a simple two-mass system, a setting that reveals the potential of the approach for locomotion, mimicking approaches based on central pattern generators. △ Less

Submitted 6 September, 2021; v1 submitted 25 March, 2021; originally announced March 2021.

A tunable mixed feedback oscillator

Authors: Weiming Che, Fulvio Forni

Abstract: The interplay of positive and negative feedback loops on different time scales appears to be a fundamental mechanisms for robust and tunable oscillations in both biological systems and electro-mechanical systems. We develop a detailed analysis of a basic three dimensional Lure model to show how controlled oscillations arise from the tuning of positive and negative feedback strengths. Our analysis… ▽ More The interplay of positive and negative feedback loops on different time scales appears to be a fundamental mechanisms for robust and tunable oscillations in both biological systems and electro-mechanical systems. We develop a detailed analysis of a basic three dimensional Lure model to show how controlled oscillations arise from the tuning of positive and negative feedback strengths. Our analysis is based on dominance theory and confirms, from a system-theoretic perspective, that the mixed feedback is a fundamental enabler of robust oscillations. Our results are not limited to three dimensional systems and extend to larger systems via passivity theory, and to uncertain systems via small gain arguments. △ Less

Submitted 5 April, 2021; v1 submitted 17 November, 2020; originally announced November 2020.

Comments: Submitted to European Control Conference 2021

AGIF: An Adaptive Graph-Interactive Framework for Joint Multiple Intent Detection and Slot Filling

Authors: Libo Qin, Xiao Xu, Wanxiang Che, Ting Liu

Abstract: In real-world scenarios, users usually have multiple intents in the same utterance. Unfortunately, most spoken language understanding (SLU) models either mainly focused on the single intent scenario, or simply incorporated an overall intent context vector for all tokens, ignoring the fine-grained multiple intents information integration for token-level slot prediction. In this paper, we propose an… ▽ More In real-world scenarios, users usually have multiple intents in the same utterance. Unfortunately, most spoken language understanding (SLU) models either mainly focused on the single intent scenario, or simply incorporated an overall intent context vector for all tokens, ignoring the fine-grained multiple intents information integration for token-level slot prediction. In this paper, we propose an Adaptive Graph-Interactive Framework (AGIF) for joint multiple intent detection and slot filling, where we introduce an intent-slot graph interaction layer to model the strong correlation between the slot and intents. Such an interaction layer is applied to each token adaptively, which has the advantage to automatically extract the relevant intents information, making a fine-grained intent information integration for the token-level slot prediction. Experimental results on three multi-intent datasets show that our framework obtains substantial improvement and achieves the state-of-the-art performance. In addition, our framework achieves new state-of-the-art performance on two single-intent datasets. △ Less

Submitted 17 October, 2020; v1 submitted 21 April, 2020; originally announced April 2020.

Comments: Accepted at Findings of EMNLP 2020. Data and code are available at this [URL] (https://github.com/LooperXX/AGIF)