arXiv subjects

Victor O. K. Li

Publications and source records attributed to Victor O. K. Li.

At least 19 recordsLinked to original sources

arXiv2026

Entropy-Guided Reverse-Causal AI to Identify Upstream Bottleneck Genes for Alzheimer's Drug Discovery

Identifying upstream regulators that connect several disease processes to therapeutic interventions is a central objective in Alzheimer's disease drug discovery. We propose an entropy-guided reverse-causal framework that makes candidate bottleneck genes the organizing link between disease mechanisms, pathways, molecular targets and drugs. The methodology integrates five stages: an Alzheimer's-specific knowledge graph with language-model assistance and expert review; reverse tracing from drugs to candidate genes; entropy-guided prioritization; forward propagation to drugs and complementary combinations; and staged validation with evidence feedback. The novelty lies in integrating upstream bottleneck identification, entropy-guided prioritization and iterative therapeutic selection within a dynamic, bidirectional discovery architecture. We demonstrate its molecular tracing and gene-prioritization components in a computational feasibility study using DeepDrug2 and MSigDB pathway annotations. Tracing amlodipine, indapamide and atorvastatin through a network of 11,300 molecular and drug nodes identifies 46 routes to nine genes. EGFR is the leading candidate, supported by 26 routes from all three drugs; MME and MAF rank next. These results show how pharmacological starting points can identify shared candidate genes with defined molecular connections. The framework's scientific significance lies in connecting convergent disease mechanisms to systematic intervention selection, with preservation of cognition and independence as the translational objective.

cs.CE↗

arXiv2025

DECT: Harnessing LLM-assisted Fine-Grained Linguistic Knowledge and Label-Switched and Label-Preserved Data Generation for Diagnosis of Alzheimer's Disease

Alzheimer's Disease (AD) is an irreversible neurodegenerative disease affecting 50 million people worldwide. Low-cost, accurate identification of key markers of AD is crucial for timely diagnosis and intervention. Language impairment is one of the earliest signs of cognitive decline, which can be used to discriminate AD patients from normal control individuals. Patient-interviewer dialogues may be used to detect such impairments, but they are often mixed with ambiguous, noisy, and irrelevant information, making the AD detection task difficult. Moreover, the limited availability of AD speech samples and variability in their speech styles pose significant challenges in developing robust speech-based AD detection models. To address these challenges, we propose DECT, a novel speech-based domain-specific approach leveraging large language models (LLMs) for fine-grained linguistic analysis and label-switched label-preserved data generation. Our study presents four novelties: We harness the summarizing capabilities of LLMs to identify and distill key Cognitive-Linguistic information from noisy speech transcripts, effectively filtering irrelevant information. We leverage the inherent linguistic knowledge of LLMs to extract linguistic markers from unstructured and heterogeneous audio transcripts. We exploit the compositional ability of LLMs to generate AD speech transcripts consisting of diverse linguistic patterns to overcome the speech data scarcity challenge and enhance the robustness of AD detection models. We use the augmented AD textual speech transcript dataset and a more fine-grained representation of AD textual speech transcript data to fine-tune the AD detection model. The results have shown that DECT demonstrates superior model performance with an 11% improvement in AD detection accuracy on the datasets from DementiaBank compared to the baselines.

cs.CL↗

arXiv2021

Show Me How To Revise: Improving Lexically Constrained Sentence Generation with XLNet

Lexically constrained sentence generation allows the incorporation of prior knowledge such as lexical constraints into the output. This technique has been applied to machine translation, and dialog response generation. Previous work usually used Markov Chain Monte Carlo (MCMC) sampling to generate lexically constrained sentences, but they randomly determined the position to be edited and the action to be taken, resulting in many invalid refinements. To overcome this challenge, we used a classifier to instruct the MCMC-based models where and how to refine the candidate sentences. First, we developed two methods to create synthetic data on which the pre-trained model is fine-tuned to obtain a reliable classifier. Next, we proposed a two-step approach, "Predict and Revise", for constrained sentence generation. During the predict step, we leveraged the classifier to compute the learned prior for the candidate sentence. During the revise step, we resorted to MCMC sampling to revise the candidate sentence by conducting a sampled action at a sampled position drawn from the learned prior. We compared our proposed models with many strong baselines on two tasks, generating sentences with lexical constraints and text infilling. Experimental results have demonstrated that our proposed model performs much better than the previous work in terms of sentence fluency and diversity. Our code and pre-trained models are available at https://github.com/NLPCode/MCMCXLNet.

cs.CL↗

arXiv2021

Deep-AIR: A Hybrid CNN-LSTM Framework for Air Quality Modeling in Metropolitan Cities

Air pollution has long been a serious environmental health challenge, especially in metropolitan cities, where air pollutant concentrations are exacerbated by the street canyon effect and high building density. Whilst accurately monitoring and forecasting air pollution are highly crucial, existing data-driven models fail to fully address the complex interaction between air pollution and urban dynamics. Our Deep-AIR, a novel hybrid deep learning framework that combines a convolutional neural network with a long short-term memory network, aims to address this gap to provide fine-grained city-wide air pollution estimation and station-wide forecast. Our proposed framework creates 1x1 convolution layers to strengthen the learning of cross-feature spatial interaction between air pollution and important urban dynamic features, particularly road density, building density/height, and street canyon effect. Using Hong Kong and Beijing as case studies, Deep-AIR achieves a higher accuracy than our baseline models. Our model attains an accuracy of 67.6%, 77.2%, and 66.1% in fine-grained hourly estimation, 1-hr, and 24-hr air pollution forecast for Hong Kong, and an accuracy of 65.0%, 75.3%, and 63.5% for Beijing. Our saliency analysis has revealed that for Hong Kong, street canyon and road density are the best estimators for NO2, while meteorology is the best estimator for PM2.5.

cs.LG↗

arXiv2021

AQEyes: Visual Analytics for Anomaly Detection and Examination of Air Quality Data

Anomaly detection plays a key role in air quality analysis by enhancing situational awareness and alerting users to potential hazards. However, existing anomaly detection approaches for air quality analysis have their own limitations regarding parameter selection (e.g., need for extensive domain knowledge), computational expense, general applicability (e.g., require labeled data), interpretability, and the efficiency of analysis. Furthermore, the poor quality of collected air quality data (inconsistently formatted and sometimes missing) also increases the difficulty of analysis substantially. In this paper, we systematically formulate design requirements for a system that can solve these limitations and then propose AQEyes, an integrated visual analytics system for efficiently monitoring, detecting, and examining anomalies in air quality data. In particular, we propose a unified end-to-end tunable machine learning pipeline that includes several data pre-processors and featurizers to deal with data quality issues. The pipeline integrates an efficient unsupervised anomaly detection method that works without the use of labeled data and overcomes the limitations of existing approaches. Further, we develop an interactive visualization system to visualize the outputs from the pipeline. The system incorporates a set of novel visualization and interaction designs, allowing analysts to visually examine air quality dynamics and anomalous events in multiple scales and from multiple facets. We demonstrate the performance of this pipeline through a quantitative evaluation and show the effectiveness of the visualization system using qualitative case studies on real-world datasets.

cs.HC↗

arXiv2020

On the Sparsity of Neural Machine Translation Models

Modern neural machine translation (NMT) models employ a large number of parameters, which leads to serious over-parameterization and typically causes the underutilization of computational resources. In response to this problem, we empirically investigate whether the redundant parameters can be reused to achieve better performance. Experiments and analyses are systematically conducted on different datasets and NMT architectures. We show that: 1) the pruned parameters can be rejuvenated to improve the baseline model by up to +0.8 BLEU points; 2) the rejuvenated parameters are reallocated to enhance the ability of modeling low-level lexical information.

cs.CL↗

arXiv2020

Max-min Fairness of K-user Cooperative Rate-Splitting in MISO Broadcast Channel with User Relaying

Cooperative Rate-Splitting (CRS) strategy, relying on linearly precoded rate-splitting at the transmitter and opportunistic transmission of the common message by the relaying user, has recently been shown to outperform typical Non-cooperative Rate-Splitting (NRS), Cooperative Non-Orthogonal Multiple Access (C-NOMA) and Space Division Multiple Access (SDMA) in a two-user Multiple Input Single Output (MISO) Broadcast Channel (BC) with user relaying. In this work, the existing two-user CRS transmission strategy is generalized to the K-user case. We study the problem of jointly optimizing the precoders, message split, time slot allocation, and relaying user scheduling with the objective of maximizing the minimum rate among users. An efficient self-organizing relaying protocol is first proposed followed by a Successive Convex Approximation (SCA)-based algorithm to jointly optimize time slot, precoders and message split. Numerical results show that the worst-case achievable rate achieved by CRS is significantly increased over that of NRS and SDMA in a wide range of network loads and user deployments. Importantly, the proposed SCA-based algorithm dramatically reduces the computational complexity without any rate loss compared with the conventional algorithm in the literature of CRS. Therefore, we conclude that the proposed K-user CRS is more powerful than the existing transmission schemes.

eess.SP↗

arXiv2020

Facial Action Unit Intensity Estimation via Semantic Correspondence Learning with Dynamic Graph Convolution

The intensity estimation of facial action units (AUs) is challenging due to subtle changes in the person's facial appearance. Previous approaches mainly rely on probabilistic models or predefined rules for modeling co-occurrence relationships among AUs, leading to limited generalization. In contrast, we present a new learning framework that automatically learns the latent relationships of AUs via establishing semantic correspondences between feature maps. In the heatmap regression-based network, feature maps preserve rich semantic information associated with AU intensities and locations. Moreover, the AU co-occurring pattern can be reflected by activating a set of feature channels, where each channel encodes a specific visual pattern of AU. This motivates us to model the correlation among feature channels, which implicitly represents the co-occurrence relationship of AU intensity levels. Specifically, we introduce a semantic correspondence convolution (SCC) module to dynamically compute the correspondences from deep and low resolution feature maps, and thus enhancing the discriminability of features. The experimental results demonstrate the effectiveness and the superior performance of our method on two benchmark datasets.

cs.CV↗

arXiv2019

Go From the General to the Particular: Multi-Domain Translation with Domain Transformation Networks

The key challenge of multi-domain translation lies in simultaneously encoding both the general knowledge shared across domains and the particular knowledge distinctive to each domain in a unified model. Previous work shows that the standard neural machine translation (NMT) model, trained on mixed-domain data, generally captures the general knowledge, but misses the domain-specific knowledge. In response to this problem, we augment NMT model with additional domain transformation networks to transform the general representations to domain-specific representations, which are subsequently fed to the NMT decoder. To guarantee the knowledge transformation, we also propose two complementary supervision signals by leveraging the power of knowledge distillation and adversarial learning. Experimental results on several language pairs, covering both balanced and unbalanced multi-domain translation, demonstrate the effectiveness and universality of the proposed approach. Encouragingly, the proposed unified model achieves comparable results with the fine-tuning approach that requires multiple models to preserve the particular knowledge. Further analyses reveal that the domain transformation networks successfully capture the domain-specific knowledge as expected.

cs.CL↗

arXiv2019

Rate-Splitting for Multi-Antenna Non-Orthogonal Unicast and Multicast Transmission: Spectral and Energy Efficiency Analysis

In a Non-Orthogonal Unicast and Multicast (NOUM) transmission system, a multicast stream intended to all the receivers is superimposed in the power domain on the unicast streams. One layer of Successive Interference Cancellation (SIC) is required at each receiver to remove the multicast stream before decoding its intended unicast stream. In this paper, we first show that a linearly-precoded 1-layer Rate-Splitting (RS) strategy at the transmitter can efficiently exploit this existing SIC receiver architecture. We further propose multi-layer transmission strategies based on the generalized RS and power-domain Non-Orthogonal Multiple Access (NOMA). Two different objectives are studied for the design of the precoders, namely, maximizing the Weighted Sum Rate (WSR) of the unicast messages and maximizing the system Energy Efficiency (EE), both subject to Quality of Service (QoS) rate requirements of all the messages and a sum power constraint. A Weighted Minimum Mean Square Error (WMMSE)-based algorithm and a Successive Convex Approximation (SCA)-based algorithm are proposed to solve the WSR and EE problems, respectively. Numerical results show that the proposed RS-assisted NOUM transmission strategies are more spectrally and energy efficient than the conventional Multi-User Linear-Precoding (MU-LP), Orthogonal Multiple Access (OMA) and power-domain NOMA in a wide range of user deployments (with a diversity of channel directions, channel strengths and qualities of channel state information at the transmitter) and network loads (underloaded and overloaded regimes). It is superior for the downlink multi-antenna NOUM transmission.

cs.IT↗

arXiv2019

Rate-Splitting for Multi-User Multi-Antenna Wireless Information and Power Transfer

In a multi-user multi-antenna Simultaneous Wireless Information and Power Transfer (SWIPT) network, the transmitter sends information to the Information Receivers (IRs) and energy to Energy Receivers (ERs) concurrently. A conventional approach is based on Multi-User Linear Precoding (MU--LP) where each IR directly decodes the intended stream by fully treating the interference from other IRs and ERs as noise. In this paper, we investigate the application of linearly-precoded Rate-Splitting (RS) in Multiple Input Single Output (MISO) SWIPT Broadcast Channel (BC). By splitting the messages of IRs into private and common parts and encoding the common parts into a common stream decoded by all IRs, RS manages the interference dynamically. The precoders are designed such that the Weighted Sum Rate (WSR) of IRs is maximized under the total transmit power constraint and the sum energy constraint for ERs. Numerical results show that the proposed RS-assisted strategy provides a better rate-energy tradeoff in MISO SWIPT BC. Under a sum energy constraint of ERs, RS-assisted strategy achieves better WSR performance of IRs than MU--LP and NOMA in a wide range of IR and ER deployments. Hence, we draw the conclusion that RS is superior for downlink SWIPT networks.

cs.IT↗

arXiv2019

Improved Zero-shot Neural Machine Translation via Ignoring Spurious Correlations

Zero-shot translation, translating between language pairs on which a Neural Machine Translation (NMT) system has never been trained, is an emergent property when training the system in multilingual settings. However, naive training for zero-shot NMT easily fails, and is sensitive to hyper-parameter setting. The performance typically lags far behind the more conventional pivot-based approach which translates twice using a third language as a pivot. In this work, we address the degeneracy problem due to capturing spurious correlations by quantitatively analyzing the mutual information between language IDs of the source and decoded sentences. Inspired by this analysis, we propose to use two simple but effective approaches: (1) decoder pre-training; (2) back-translation. These methods show significant improvement (4~22 BLEU points) over the vanilla zero-shot translation on three challenging multilingual datasets, and achieve similar or better results than the pivot-based approach.

cs.CL↗

arXiv2019

Rate-Splitting Multiple Access for Coordinated Multi-Point Joint Transmission

As a promising downlink multiple access scheme, Rate-Splitting Multiple Access (RSMA) has been shown to achieve superior spectral and energy efficiencies compared with Space-Division Multiple Access (SDMA) and Non-Orthogonal Multiple Access (NOMA) in downlink single-cell systems. By relying on linearly precoded rate-splitting at the transmitter and successive interference cancellation at the receivers, RSMA has the capability of partially decoding the interference and partially treating the interference as noise, and therefore copes with a wide range of user deployments and network loads. In this work, we further study RSMA in downlink Coordinated Multi-Point (CoMP) Joint Transmission (JT) networks by investigating the optimal beamformer design to maximize the Weighted Sum-Rate (WSR) of all users subject to individual Quality of Service (QoS) rate constraints and per base station power constraints. Numerical results show that, in CoMP JT, RSMA achieves significant WSR improvement over SDMA and NOMA in a wide range of inter-user and inter-cell channel strength disparities. Specifically, SDMA (resp. NOMA) is more suited to deployments with little (resp. large) inter-user channel strength disparity and large (resp. little) inter-cell channel disparity, while RSMA is suited to any deployment. We conclude that RSMA provides rate, robustness and QoS enhancements over SDMA and NOMA in CoMP JT networks.

cs.IT↗

arXiv2018

Energy Efficiency of Rate-Splitting Multiple Access, and Performance Benefits over SDMA and NOMA

Rate-Splitting Multiple Access (RSMA) is a general and powerful multiple access framework for downlink multi-antenna systems, and contains Space-Division Multiple Access (SDMA) and Non-Orthogonal Multiple Access (NOMA) as special cases. RSMA relies on linearly precoded rate-splitting with Successive Interference Cancellation (SIC) to decode part of the interference and treat the remaining part of the interference as noise. Recently, RSMA has been shown to outperform both SDMA and NOMA rate-wise in a wide range of network loads (underloaded and overloaded regimes) and user deployments (with a diversity of channel directions, channel strengths and qualities of Channel State Information at the Transmitter). Moreover, RSMA was shown to provide spectral efficiency and QoS enhancements over NOMA at a lower computational complexity for the transmit scheduler and the receivers. In this paper, we build upon those results and investigate the energy efficiency of RSMA compared to SDMA and NOMA. Considering a multiple-input single-output broadcast channel, we show that RSMA is more energy-efficient than SDMA and NOMA in a wide range of user deployments (with a diversity of channel directions and channel strengths). We conclude that RSMA is more spectrally and energy-efficient than SDMA and NOMA.

cs.IT↗

arXiv2018

Large Margin Few-Shot Learning

The key issue of few-shot learning is learning to generalize. This paper proposes a large margin principle to improve the generalization capacity of metric based methods for few-shot learning. To realize it, we develop a unified framework to learn a more discriminative metric space by augmenting the classification loss function with a large margin distance loss function for training. Extensive experiments on two state-of-the-art few-shot learning methods, graph neural networks and prototypical networks, show that our method can improve the performance of existing models substantially with very little computational overhead, demonstrating the effectiveness of the large margin principle and the potential of our method.

cs.LG↗

arXiv2018

A Stable and Effective Learning Strategy for Trainable Greedy Decoding

Beam search is a widely used approximate search strategy for neural network decoders, and it generally outperforms simple greedy decoding on tasks like machine translation. However, this improvement comes at substantial computational cost. In this paper, we propose a flexible new method that allows us to reap nearly the full benefits of beam search with nearly no additional computational cost. The method revolves around a small neural network actor that is trained to observe and manipulate the hidden state of a previously-trained decoder. To train this actor network, we introduce the use of a pseudo-parallel corpus built using the output of beam search on a base model, ranked by a target quality metric like BLEU. Our method is inspired by earlier work on this problem, but requires no reinforcement learning, and can be trained reliably on a range of models. Experiments on three parallel corpora and three architectures show that the method yields substantial improvements in translation quality and speed over each base system.

cs.CL↗

arXiv2018

Meta-Learning for Low-Resource Neural Machine Translation

In this paper, we propose to extend the recently introduced model-agnostic meta-learning algorithm (MAML) for low-resource neural machine translation (NMT). We frame low-resource translation as a meta-learning problem, and we learn to adapt to low-resource languages based on multilingual high-resource language tasks. We use the universal lexical representation~\citep{gu2018universal} to overcome the input-output mismatch across different languages. We evaluate the proposed meta-learning strategy using eighteen European languages (Bg, Cs, Da, De, El, Es, Et, Fr, Hu, It, Lt, Nl, Pl, Pt, Sk, Sl, Sv and Ru) as source tasks and five diverse languages (Ro, Lv, Fi, Tr and Ko) as target tasks. We show that the proposed approach significantly outperforms the multilingual, transfer learning based approach~\citep{zoph2016transfer} and enables us to train a competitive NMT system with only a fraction of training examples. For instance, the proposed approach can achieve as high as 22.04 BLEU on Romanian-English WMT'16 by seeing only 16,000 translated words (~600 parallel sentences).

cs.CL↗

arXiv2018

Multi-Region Ensemble Convolutional Neural Network for Facial Expression Recognition

Facial expressions play an important role in conveying the emotional states of human beings. Recently, deep learning approaches have been applied to image recognition field due to the discriminative power of Convolutional Neural Network (CNN). In this paper, we first propose a novel Multi-Region Ensemble CNN (MRE-CNN) framework for facial expression recognition, which aims to enhance the learning power of CNN models by capturing both the global and the local features from multiple human face sub-regions. Second, the weighted prediction scores from each sub-network are aggregated to produce the final prediction of high accuracy. Third, we investigate the effects of different sub-regions of the whole face on facial expression recognition. Our proposed method is evaluated based on two well-known publicly available facial expression databases: AFEW 7.0 and RAF-DB, and has been shown to achieve the state-of-the-art recognition accuracy.

cs.CV↗