Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks270 from public data
- 364
- Federated machine learning in healthcare: A systematic review on clinical applications and technical architecture280
The need to address the barriers to clinical translation and to assess the real-world impact of Federated learning in this new digital data-driven healthcare scene is highlighted.
- 242
- 206
- Medical Image Segmentation using Squeeze-and-Expansion Transformers201
Segtran is an alternative segmentation framework based on transformers, which has unlimited effective receptive fields even at high feature resolutions, and a new positional encoding scheme for transformers is proposed, imposing a continuity inductive bias for images.
- Federated benchmarking of medical artificial intelligence with MedPerf200
The current challenges healthcare and AI communities face, the need for an open platform, the design philosophy of MedPerf, its current implementation status and real-world deployment, the authors' roadmap and, importantly, the use of Med perf with multiple international institutions within cloud-based technology and on-premises scenarios are described.
- Retinal photograph-based deep learning algorithms for myopia and a blockchain platform to facilitate artificial intelligence medical research: a retrospective multicohort study163
Deep learning algorithms can be effective tools for risk stratification and screening of myopic macular degeneration and high myopia among the large global population with myopia.
- CRAFT: Cross-Attentional Flow Transformer for Robust Optical Flow149
This work proposes a new architecture “CRoss-Attentional Flow Trans-former” (CRAFT), aiming to revitalize the correlation volume computation, and designed an image shifting attack that shifts input images to generate large artificial motions.
- 128
- Two-Phase Multi-Party Computation Enabled Privacy-Preserving Federated Learning121
This paper proposes to adopt Multi-Party Computation (MPC) to achieve privacy-preserving model aggregation for FL, and proposes to develop a two-phase mechanism by electing a small committee and providing MPC-enabled model aggregation service to a larger number of participants through the committee.
- Natural Language Video Localization: A Revisit in Span-based Question Answering Framework116
This study suggests that the span-based QA framework is an effective strategy to solve the NLVL problem, and proposes a video span localizing network (VSLNet) and VSLNet-L, which address the issue of performance degradation on long videos.
- Dual Adversarial Neural Transfer for Low-Resource Named Entity Recognition114
Two variants of DATNet are investigated to explore effective feature fusion between high and low resource, and a novel Generalized Resource-Adversarial Discriminator (GRAD) is proposed to address the noisy and imbalanced training data.
- 112
- Adversarial multimodal fusion with attention mechanism for skin lesion classification using clinical and dermoscopic images103
A novel method, named adversarial multimodal fusion with attention mechanism (AMFAM), to perform multi-task skin lesion classification that outperforms the current state-of-the-art methods and improves the average AUC score by above 2% on the test set.
- 99
- 98
- The Singapore National Precision Medicine Strategy95
Singapore’s efforts to implement a National Precision Medicine Strategy through the integration of genomic, clinical and lifestyle data of up to one million Singaporean individuals are discussed.
- An Aggregation-Free Federated Learning for Tackling Data Heterogeneity93
FedAF inherently avoids the issue of client drift, enhances the quality of condensed data amid notable data heterogeneity, and improves the global model performance, leading to superior global model accuracy and faster convergence.
- 93
- IMPLEMENTATION OF THE FDTD METHOD BASED ON LORENTZ-DRUDE DISPERSIVE MODEL ON GPU FOR PLASMONICS APPLICATIONS91
This work presents a three-dimensional flnite difierence time domain method on graphics processing unit (GPU) for plasmonics applications and demonstrates that the implementation of the FDTD method on GPU ofiers signiflcant speed up as compared to the traditional CPU implementations.
- Sentence-level Prompts Benefit Composed Image Retrieval89
This work demonstrates that learning an appropriate sentence-level prompt for the relative caption (SPRC) is sufficient for achieving effective composed image retrieval and proposes to leverage pretrained V-L models, e.g., BLIP-2, to generate sentence- level prompts.
- Uncertainty-inspired open set learning for retinal anomaly identification87
UIOS provides a robust method for real-world screening of retinal anomalies and correctly predicts high uncertainty scores, which would prompt the need for a manual check in the datasets of non-target categories retinal diseases, low-quality fundus images, and non-fundus images.
- 69
- A System-Level Simulator for RRAM-Based Neuromorphic Computing Chips63
A cycle-accurate and scalable system-level simulator that can be used to study the effects of using RRAM devices in neuromorphic computing chips and shows that RTN and write variability can have adverse effects on an application, but can be mitigated through proper design choices and the implementation of a write-verify scheme.
- Multi-Instance Multi-Scale CNN for Medical Image Classification62
A Multi-Instance Multi-Scale (MIMS) CNN is proposed, which extracts patterns of different receptive fields with a shared set of convolutional kernels so that scale-invariant patterns are captured by this compact set of kernels.
- SC2Net: Sparse LSTMs for Sparse Coding61
This work recasts ISTA as a recurrent neural network unit and shows its connection with the well-known long short term memory (LSTM) model, and presents a neural network to achieve sparse codes in an end-to-end manner.
- 61
- Privacy-Preserving Weighted Federated Learning Within the Secret Sharing Framework60
It is shown that FL is a subset of MPC from the m-ary functionality point of view, and the proposed implementation is secure against honest-but-curious adversary assuming that the underlying El Gamal encryption is semantically secure.
- Detecting visually significant cataract using retinal photograph-based deep learning56
The development and validation of a retinal photograph-based, deep-learning algorithm for automated detection of visually significant cataracts, using more than 25,000 images from population-based studies are reported, report that it detects cataract with similar accuracy to ophthalmologists.
- Ladder queue55
Numerical simulations ranging from 100 to 10 million events affirm the O(1) property of LadderQ and that it is a superior structure for large-scale discrete event simulation.
- How Interpretable are Reasoning Explanations from Prompting Large Language Models?54
A simple interpretability alignment technique is introduced, termed Self-Entailment-Alignment Chain-of-thought, that yields more than 70\% improvements across multiple dimensions of interpretability.
- Optimizing the MapReduce framework on Intel Xeon Phi coprocessor53
This work develops the first MapReduce framework on the recently released Intel Xeon Phi coprocessor to take advantage of the SIMD vector processing units and proposes a vectorization friendly technique to assist the auto-vectorization as well as develop SIMD hash computation algorithms.
- Scalable framework for mapping streaming applications onto multi-GPU systems52
An efficient and scalable code generation framework that can map general purpose streaming applications onto a multi-GPU system and is implemented as a back-end of the StreamIt programming language compiler.
- 49
- 48
- Referral for disease-related visual impairment using retinal photograph-based deep learning: a proof-of-concept, model development study47
This proof-of-concept study shows the potential of a single-modality, function-focused tool in identifying visual impairment related to major eye diseases, providing more timely and pinpointed referral of patients with disease-related visual impairment from the community to tertiary eye hospitals.
- E3NE: An End-to-End Framework for Accelerating Spiking Neural Networks with Emerging Neural Encoding on FPGAs45
This end-to-end framework E3NE automates the generation of efficient SNN inference logic for FPGAs and applies various optimizations and assesses trade-offs inherent to spike-based accelerators, resulting in an efficiency superior to previous SNN hardware implementations.
- 45
- 45
- 42
- 41
- 40
- Trajectories of reported sleep duration associate with early childhood cognitive development36
Longer and more consistent night- and total sleep trajectories, and a short day sleep trajectory in early childhood were associated with better cognition at 2 and 4.5 years.
- Development and testing of a multi-lingual Natural Language Processing-based deep learning system in 10 languages for COVID-19 pandemic crisis: A multi-center study33
A multi-lingual NLP-based conversational AI chatbot, DR-COVID, which responds accurately to open-ended, COVID-19 related questions and generated answers more accurately and quickly than other chatbots is developed to facilitate pandemic education and healthcare delivery.
- Efficient Spiking Neural Networks With Radix Encoding33
This article proposes a radix-encoded SNN, which has ultrashort spike trains and develops a method to fit the radix encoding technique into the ANN-to-SNN conversion approach so that it can train radix-encoded SNNs more efficiently on mature platforms and hardware.
- 32
- 32
- Achieving Green AI with Energy-Efficient Deep Learning Using Neuromorphic Computing31
A simulator of a neuromorphic chip with up to around 20,000 neural cores that was tested to run on 512 Nvidia A100 GPUs and an FPGA-based hardware emulator for the neuro-morphic chip were developed.
- Contrastive domain adaptation with consistency match for automated pneumonia diagnosis31
This work proposes a novel method called Contrastive Domain Adaptation with Consistency Match (CDACM), which outperforms several state-of-the-art unsupervised domain adaptation approaches, and verifies the effectiveness of CDACM for automated pneumonia diagnosis using chest X-ray imaging.
- 31
- 30
- 30
- 30
- DeepFire2: A Convolutional Spiking Neural Network Accelerator on FPGAs29
DeepFire2 introduces a hardware architecture which can map large network layers efficiently across multiple super logic regions in a multi-die FPGA, and is able to deploy a large ImageNet model, while maintaining a throughput of over 1500 frames per second.
- Variations in longitudinal sleep duration trajectories from infancy to early childhood29
This is the first study to describe multiple sleep trajectories in Singaporean children and identify between-individual variability within the trajectory groups, and short, moderate, and long trajectories differed significantly in duration.
- Few-Shot Domain Adaptation with Polymorphic Transformers28
A Polymorphic Transformer (Polyformer), which can be incorporated into any DNN backbones for few-shot domain adaptation, and which can perform robustly on the target domain after being trained on a few annotated images.
- Neural Modeling of Buying Behaviour for E-Commerce from Clicking Patterns28
This paper provides a direct method in modeling the buying pattern in a clicking session by simply using the time-stamp of the clicks and shows that the result is comparable to using more massive feature engineering that requires session summarizing.
- Leveraging social networking sites for disease surveillance and public sensing: the case of the 2013 avian influenza A(H7N9) outbreak in China28
The use of a popular Chinese social networking and microblogging site, Sina Weibo, to monitor an avian influenza A(H7N9) outbreak in China and the potential for social networking sites to be used by public health agencies to enhance traditional communicable disease surveillance systems for the global surveillance of overseas public health threats is demonstrated.
- Benchmarking Quantum(-Inspired) Annealing Hardware on Practical Use Cases27
Experiments suggest that both quantum(-inspired) annealers are effective on problems with small size and simple settings, but lose their utility when facing problems in practical size and settings, and decomposition methods extend the scalability of annesalers, but they are still far away from practical use.
- 27
- 26
- 25
- 25
- An effective blockchain-based, decentralized application for smart building system management24
A blockchain-based, DApp named Uranus for smart building system management with private Ethereum blockchain, Raspberry Pi, Blynk platform, and sensors, and the blockchain technology in Uranus provides three benefits: it is tamper-resistant, it has no single point of failure, and it is trusted and auditable.
- 24
- Privacy-Preserving Technology Using Federated Learning and Blockchain in Protecting against Adversarial Attacks for Retinal Imaging23
The proposed FL algorithm overcomes the shortcoming of the traditional FL in non i.i.d. situations and remains robust to against adversarial attacks and the addition of blockchain adds further security during the transfer of model updates.
- 23
- 23
- 22
- Adversarial Semantic Hallucination for Domain Generalized Semantic Segmentation21
This work proposes an adversarial semantic hallucination approach (ASH), which combines a class-conditioned hallucination module and a semantic segmentation module to generate affine transformation parameters from semantic information in the segmentation probability maps of the source domain image.
- 21
- 20
- 19
- Learning Prompt with Distribution-Based Feature Replay for Few-Shot Class-Incremental Learning18
This paper solved FSCIL by leveraging the Vision-Language model and proposed a simple yet effective framework, named Learning Prompt with Distribution-based Feature Replay (LP-DiF), observing that simply using CLIP for zero-shot evaluation can substantially outperform the most influential methods.
- 18
- 17
- Federated Uncertainty-Aware Aggregation for Fundus Diabetic Retinopathy Staging17
This work proposes a novel federated uncertainty-aware aggregation paradigm (FedUAA), which considers the reliability of each client and produces a confidence estimation for the DR staging, and provides a robust and reliable solution for the deployment of DR diagnosis models in real-world clinical scenarios.
- 17
- 17
- 16
- 16
- 16
- Exploiting Sparsity to Accelerate Fully Connected Layers of CNN-Based Applications on Mobile SoCs16
Customized versions of the sparse matrix multiplication algorithm are proposed to speed up inference on mobile devices and make it more energy efficient and bit-representation-based algorithms that exploit sparsity to accelerate the fully connected layers of a network on the NVIDIA Jetson TK1 platform are proposed.
- 16
- 16
- Tulipse: A Visualization Framework for User-Guided Parallelization16
A framework is proposed that enables the programmer to visualize information critical for semi-automated parallelization and offers a program structure view that is augmented with key performance information, and a loop-nest dependency view that can be used to visualize data dependencies gathered from static or dynamic analyses.
- 16
- Towards Reliable Medical Image Segmentation by Modeling Evidential Calibrated Uncertainty15
A foundation model named EvidenceCap is proposed, which makes the box transparent in a quantifiable way by uncertainty estimation, and enhances the reliability, robustness, and computational efficiency of MIS.
- Rethinking Client Drift in Federated Learning: A Logit Perspective15
A new algorithm, named FedCSD, a Class prototype Similarity Distillation in a federated framework to align the local and global models to enhance the quality of global logits, and adopts an adaptive mask to filter out the terrible soft labels of the global models, thereby preventing them to mislead local optimization.
- 15
- Equality of public transit connectivity: the influence of mass rapid transit services on individual buildings for Singapore15
A graph-based Public Transit Connectivity (PTC) index is proposed to measure the accessibility of individual buildings focusing on transportation factors to evaluate the influence of existing and future Mass Rapid Transit services in Singapore as a case study.
- 15
- 14
- mHealth App to Facilitate Remote Care for Patients With COVID-19: Rapid Development of the DrCovid+ App14
The rapid development and implementation of DrCovid+ allowed for timely clinical care management for patients with COVID-19 and facilitated early patient hospital discharge and continuity of care while addressing issues relating to data security and labor-, time-, and cost-effectiveness.
- RCT: Resource Constrained Training for Edge AI14
This work proposes resource constrained training (RCT), which only keeps a quantized model throughout the training so that the memory requirement for model parameters in training is reduced, and adjusts per-layer bitwidth dynamically to save energy when a model can learn effectively with lower precision.
- 14
- 14
- 13
- Adaptive Resource Provisioning Mechanism in VEEs for Improving Performance of HLA-Based Simulations13
In order to speed up simulation execution, an Adaptive Resource Provisioning Mechanism in Virtual Execution Environments (ArmVee) is proposed, which is composed of a performance monitor and a resource manager.
- Federated Pseudo Modality Generation for Incomplete Multi-Modal MRI Reconstruction12
This paper proposes a novel communication-efficient federated learning framework (namely Fed-PMG) to address the missing modality challenge in federated multi-modal MRI reconstruction and demonstrates that the proposed method can outperform state-of-the-art methods and reach a performance similar to the ideal scenario.
- 12
- A five-safes approach to a secure and scalable genomics data repository12
RAPTOR by the Genome Institute of Singapore is a cloud-native genomics data repository and analytics platform that implements a “five-safes” framework to provide security and governance controls to data contributors and users, leveraging CSP for sharing and analysis of genomic datasets without the risk of security breaches or running afoul of regulations.
- Fast Recovery MapReduce (FAR-MR) to accelerate failure recovery in big data applications12
The performance evaluation has shown that the proposed FAR-MR can improve computing job performance by up to 62% and 45% compared to Hadoop MapReduce in the case of task failure recovery and node failure recovery, respectively.
- 12
- Table-Lookup MAC: Scalable Processing of Quantised Neural Networks in FPGA Soft Logic11
This paper introduces Table Lookup Multiply-Accumulate (TLMAC) as a framework to compile and optimise quantised neural networks for scalable lookup-based processing and demonstrates that TLMAC significantly improves the scalability of previous related works.
- 11
- 10
- Detection of Center-Involved Diabetic Macular Edema With Visual Impairment Using Multimodal Artificial Intelligence Algorithms10
The AI models showed good diagnostic performance for detection of CI-DME with visual impairment, and the multimodal (CFP and OCT) model did not offer additional benefit over the OCT model alone.
- $AiRacleX$: Automated Detection of Price Oracle Manipulations via LLM-Driven Knowledge Mining and Prompt Generation10
A novel LLM-driven framework that enables pre-deployment detection of price oracle manipulation vulnerabilities by leveraging the complementary strengths of multiple large language models (LLMs), and demonstrates strong extensibility and efficiency.
- Enabling Energy-Efficient Deployment of Large Language Models on Memristor Crossbar: A Synergy of Large and Small10
A novel architecture for the memristor crossbar is presented that enables the deployment of state-of-the-art LLM on a single chip or package, eliminating the energy and time inefficiencies associated with off-chip communication.
- 10
- Early dengue outbreak detection modeling based on dengue incidences in Singapore during 2012 to 201710
This investigation shows that the proposed two‐step framework is able to give persistent signals at the early stage of the outbreaks in 2013, 2014, and 2016, which provides early alerts of outbreaks and wins time for the early interventions and the preparation of necessary public health resources.
- MarineMAS: A multi-agent framework to aid design, modelling, and evaluation of autonomous shipping systems10
A multi-agent system (MAS) framework, called “MarineMAS,” is proposed to aid the autonomous shipping modelling and evaluation, focusing on critical knowledge transfer to make transition towards autonomous shipping, the core components to achieve system intelligence, and the potential technological set and experience gained to support efficient MAS modelling.
- Efficient Query Processing on Many-core Architectures10
In PhiDB, an OLAP query processor with simultaneous multi-threading (SMT) capabilities on Xeon Phi, a heuristic algorithm is designed to schedule the concurrent execution of query operators for better performance, to demonstrate the performance impact of Xeon Phi aware optimizations.
- 10
- 10
- Text to Image for Multi-Label Image Recognition With Joint Prompt-Adapter Learning9
T2I-PAL offers significant advantages: it eliminates the need for fully semantically annotated training images, thereby reducing the manual annotation workload, and it preserves the intrinsic mode of the CLIP model, allowing for seamless integration with any existing CLIP framework.
- Class Balance Matters to Active Class-Incremental Learning9
The Active Class-Incremental Learning (ACIL) is introduced, to select the most informative samples from the unlabeled pool to effectively train an incremental learner, aiming to maximize the performance of the resulting model.
- A Resource-efficient Spiking Neural Network Accelerator Supporting Emerging Neural Encoding9
This work presents a novel hardware architecture that can efficiently support SNN with emerging neural encoding, and is the first work to deploy the large neural network model VGG on physical FPGA-based neuromorphic hardware.
- A Network‐Based Impact Measure for Propagated Losses in a Supply Chain Network Consisting of Resilient Components9
This work forms a network-based measure of the impact of a disruption loss in a supply chain propagating downstream from an originating node that takes into account the loss profile of the originating node, the structure of the supply network, and the resilience of the network components.
- 9
- 9
- 9
- Diffusion-Enhanced Test-Time Adaptation with Text and Image Augmentation8
A novel test-time adaptation method that utilizes a pre-trained generative model for multi-modal augmentation of each test sample from unknown new domains by combining augmented data from pre-trained vision and language models is introduced.
- A New Perspective to Boost Performance Fairness For Medical Federated Learning8
Fed-LWR is proposed to improve performance fairness from the perspective of feature shift, a key issue influencing the performance of medical FL systems caused by domain shift, and dynamically perceive the bias of the global model across all hospitals.
- 8
- 8
- An integrated language-vision foundation model for conversational diagnostics and triaging in primary eye care7
Through self-supervised learning and a user-friendly platform, Meta-EyeFM addresses the scarcity of skilled eye care professionals, offering scalable, explainable AI for enhancing vision screening and disease triage globally.
- Is quantum optimization ready? An effort towards neural network compression using adiabatic quantum computing7
Experiments demonstrate that adiabatic quantum computing (AQC) not only outperforms classical algorithms like genetic algorithms and reinforcement learning in terms of time efficiency but also excels at identifying global optima.
- Semi-rPPG: Semi-Supervised Remote Physiological Measurement With Curriculum Pseudo-Labeling7
A novel semi-supervised learning (SSL) method named semi-rPPG that combines curriculum pseudo-labeling and consistency regularization is proposed to extract intrinsic physiological features from unlabeled data without impairing the model from noises.
- 7
- 7
- 7
- 7
- 7
- 7
- 7
- A network perspective on the calamity, induced inaccessibility of communities and the robustness of centralized, landbound relief efforts7
The robustness of centralized, landbound relief operations' capability to promptly reach areas affected by a disaster event from a network perspective is examined, showing that, from a node designated as the center for relief operations (a "relief center"), damage to a road network causes a substantial fraction of the other nodes to become initially inaccessible from any relief effort.
- 7
- 7
- 6
- 6
- VQA4CIR: Boosting Composed Image Retrieval with Visual Question Answering6
This work provides a Visual Question Answering (VQA) perspective to boost the performance of CIR and proposes a post-processing approach that outperforms state-of-the-art CIR methods on the CIRR and Fashion-IQ datasets.
- 6
- From Generalist to Specialist: Adapting Vision Language Models via Task-Specific Visual Instruction Tuning6
VITask is introduced, a novel framework that enhances task-specific adaptability of VLMs by integrating task-specific models (TSMs) and offers practical advantages such as flexible TSM integration and robustness to incomplete instructions, making it a versatile and efficient solution for task-specific VLM tuning.
- 6
- NCPower: Power Modelling for NVM-based Neuromorphic Chip6
This work systemically developed analytical models based on physical laws, and integrated NCPower, an energy consumption estimator for NVM-based neuromorphic chip, and analyzed the accuracy and energy consumption of both the traditional multi-spike based SNN and the new single- Spiking neural networks.
- 6
- 6
- Understanding Natural Disasters as Risks in Supply Chain Management through Web Data Analysis6
A real-time data crawler is developed to collect and analyze tweets that are identified as relevant to natural disasters and a visualization platform customizable to users’ requirements is implemented as part of the decision making dashboard for supply chain risk management.
- 6
- Multi-Comparison of Different Ocular Imaging Modality-based Deep Learning Models for Visually Significant Cataract Detection5
The retinal model is highlighted as a promising tool for detecting VSC, outperforming slit beam and diffuse anterior segment models, and could enable opportunistic cataract screening with minimal add-on cost.
- 5
- 5
- Associations between sleep trajectories up to 54 months and cognitive school readiness in 4 year old preschool children5
Findings suggest that individual differences in longitudinal sleep duration patterns from as early as 3 months of age may be associated with language and numeracy aspects of school readiness at 48–50 ages, as early school readiness, particularly the domains of language and mathematics, is a key predictor of subsequent academic achievement.
- Deep Neural Network Augments Performance of Junior Residents in Diagnosing COVID-19 Pneumonia on Chest Radiographs5
While the AI model improved junior residents’ performance, a decline in performance was observed on the external test compared to the internal test set, suggesting a domain shift between the patient dataset and the external dataset, highlighting the need for future research on test-time training domain adaptation to address this issue.
- 5
- 5
- 5
- 4
- 4
- Reliable Federated Disentangling Network for Non-IID Domain Feature4
RfedDis is the FL approach to combine evidential uncertainty with feature disentangling, enhancing both performance and reliability in handling non-IID domain features, providing outstanding performance coupled with a high degree of reliability.
- CPT: Consistent Proxy Tuning for Black-box Optimization4
Different from Proxy-tuning, CPT additionally exploits the frozen large black-box model and another frozen small white-box model, ensuring consistency between training-stage optimization objective and test-time proxies, which benefits Proxy-tuning and enhances model performance.
- 4
- 4
- 4
- Risk Analysis and Quantification of Vulnerability in Maritime Transportation Network Using AIS Data4
The auto identification system (AIS) data that provides us with the real-time location of vessels is analyzed and a method to compute the vulnerability and importance analytically is introduced.
- Efficient analysis of mode profiles in elliptical microcavity using dynamic-thermal electron-quantum medium FDTD method4
The dynamic-thermal electron-quantum medium finite-difference time-domain (DTEQM-FDTD) method is used for efficient analysis of mode profile in elliptical microcavity and it is observed that at some length ratios, cavity mode is excited instead of whispering gallery mode, depicting that mode profiles are length ratio dependent.
- 4
- 4
- 4
- 3
- Improving Learning of New Diseases Through Knowledge-Enhanced Initialization for Federated Adapter Tuning3
This work introduces Federated Knowledge-Enhanced Initialization (FedKEI), a novel framework that leverages cross-client and cross-task transfer from past knowledge to generate informed initializations for learning new tasks with adapters.
- 3
- 3
- Optimizing for In-Memory Deep Learning With Emerging Memory Technology3
This article proposes three optimization techniques that can improve the accuracy of the in-memory deep learning model while maximizing its energy efficiency and shows that this solution can fully recover most models’ state-of-the-art (SOTA) accuracy and achieves at least an order of magnitude higher energy efficiency than the SOTA.
- Author Correction: Detecting visually significant cataract using retinal photograph-based deep learning3
Yih-Chung Tham, Jocelyn Hui Lin Goh, Ayesha Anees, Xiaofeng Lei, Tyler Hyungtaek Rim, Miao-Li Chee, Ya Xing Wang, Jost B. Jonas, Rahat Husain, Charumathi Sabanayagam, Jie Jin Wang, Qingyu Chen, Zhiyong Lu, Tiarnan D.
- Reliable Joint Segmentation of Retinal Edema Lesions in OCT Images3
A novel reliable multi-scale wavelet-enhanced transformer network, which can provide accurate segmentation results with reliability assessment and a novel uncertainty segmentation head based on the subjective logical evidential theory is introduced to generate the final segmentsation results.
- 3
- Deep N-ary Error Correcting Output Codes3
To facilitate the training of N-ary ECOC with deep learning base learners, this work proposes three different variants of parameter sharing architectures for deep N-ARY ECOC, and demonstrates its generalization ability.
- 3
- Multi-discriminator Generative Adversarial Networks for Improved Thin Retinal Vessel Segmentation3
A novel multiscale segmentation method named Multiple discriminator generative adversarial network (MuGAN), which contains multiple discriminators with different effective receptive fields, which are sensitive to features at different scales.
- Automated Hyper-parameter Tuning for Machine Learning Models in Machine Health Prognostics3
This paper considers the use of Bayesian optimization algorithms, which automate an effective choice of HP-config by solving the associated hyperparameter optimization problem.
- IMPLEMENTATION OF THE LORENTZ–DRUDE MODEL INCORPORATED FDTD METHOD ON MULTIPLE GPUs FOR PLASMONICS APPLICATIONS3
It is shown that by using multiple GPUs in parallel fashion, significant reduction in the simulation time can be achieved as compared to the single GPU.
- 3
- A Network Connectivity Embedded Clustering Approach for Supply Chain Risk Assessment3
A network connectivity embedded k-means clustering approach has been proposed to determine at-risk clusters of nodes which share similar risk profiles and linkages with the focal company.
- 3
- 3
- A case study on dynamic kernel adaptation in a component-based infectious disease simulator3
A generalized component-based optimizer for the compositional adaptation of the compute-intensive kernels and possesses dynamic adaptation feature which selects the most appropriate kernel implementation based on runtime conditions so as to achieve performance that is otherwise unattainable with a single implementation.
- MerMED-FM: Multimodal, Multi-Disease Medical Imaging Foundation Model2
Multimodal, Multi-Disease Medical Imaging Foundation Model MerMED-FM has the potential to be a highly adaptable, versatile, cross-specialty foundation model that enables robust interpretation of medical imaging across diverse medical disciplines.
- Are Traditional Deep Learning Model Approaches as Effective as a Retinal-Specific Foundation Model for Ocular and Systemic Disease Detection?2
Traditional DL models are mostly comparable to RETFound for ocular disease detection with large datasets, however, RETFound is superior in systemic disease detection with smaller datasets.
- Multimodal, Multi-Disease Medical Imaging Foundation Model (MerMED-FM)2
MerMED-FM has the potential to be a highly adaptable, versatile, cross-specialty foundation model that enables robust medical imaging interpretation across diverse medical disciplines.
- 2
- EVLF-FM: Explainable Vision Language Foundation Model for Medicine2
EVLF-FM is an early multi-disease VLM model with explainability and reasoning capabilities that could advance adoption of and trust in foundation models for real-world clinical deployment.
- BForTFin: A Financial Domain-Aware Multiscale Evaluation Method for Time-Series Foundation Models2
A surprising finding is that Lag-Llama, the smallest model evaluated, consistently delivers superior accuracy in several scenarios, particularly at higher frequencies, challenging the assumption that larger models inherently perform better.
- Prediction of the risk of diabetic foot from corneal nerve images using deep learning algorithms2
The authors' DLAs integrating corneal nerve images with HbA1c or serum creatinine have good performance in predicting and stratifying diabetic foot risk, providing a new screening approach.
- 2
- 2
- Partially Supervised Unpaired Multi-modal Learning for Label-Efficient Medical Image Segmentation2
This paper investigates the use of partially labeled data for label-efficient unpaired multi-modal learning and proposes a novel Decomposed partial class adaptation with snapshot Ensembled Self-Training (DEST) framework for it, which outperforms existing methods significantly.
- 2
- 2
- Localizing Anatomical Landmarks in Ocular Images Using Zoom-In Attentive Networks2
A zoom-in attentive network (ZIAN) for anatomical landmark localization in ocular images achieves promising performances and outperforms state-of-the-art localization methods.
- 2
- EDCompress: Energy-Aware Model Compression with Dataflow.2
EDCompress is proposed, an Energy-aware model compression method for various Dataflows that can effectively reduce the energy consumption of various edge devices, with different dataflow types, and find the optimal dataflow type for specific neural networks in terms of energy consumption.
- 2
- 2
- 2
- 2
- A k-means clustering for supply chain risk management with embedded network connectivity2
A network connectivity embedded k-means clustering approach has been proposed to determine at-risk clusters of nodes that share similar risk profiles and linkages with the focal company.
- Transparent three-phase Byzantine fault tolerance for parallel and distributed simulations2
A three-phase Byzantine Fault Tolerance mechanism is proposed based on a transparent middleware approach and the replication, checkpointing and message logging techniques are integrated in the mechanism for the purpose of enhancing simulation performance and reducing fault tolerance cost.
- 2
- 2
- Notice of Violation of IEEE Publication Principles: Scientific Workflow Partitioning and Data Flow Optimization in Hybrid Clouds2
A novel approach to refine workflow structure and optimize intermediate data transfers without changing the scale of scheduled cloud service for large-scale scientific workflows containing thousands (or even millions) of tasks is described.
- A new centrality measure for probabilistic diffusion in network2
This paper proposes a new centrality measure, the infection diffusion eigenvector centrality (IDEC), which considers all eigenvalues and eigenvectors and conducts the recovery probability enforcement simulation, which indicates that the IDEC shows better predictability than other centrality measures when the effective infection ratio is low.
- 2
- Component-based design for adaptive large-scale infectious disease simulation2
The usefulness of componentbased approach in the development of an infectious disease simulator is demonstrated by exploring the possibility of self performance tuning at runtime through the use of hot-swappable components by incrementally develop optimised component variants easily.
- 2
- Twol-amalgamated priority queues2
Detailed empirical results show that the Twol-amalgamated priority queues consistently outperform those basal structures and are suitable for implementation in sizeable application scenarios such as, but not limited to, large-scale discrete event simulation.
- 1
- Optimizing Neural Networks with Learnable Non-Linear Activation Functions via Lookup-Based FPGA Acceleration1
Evaluations using KANs demonstrate that the FPGA-based design achieves superior computational speed and over 104 times higher energy efficiency compared to edge CPUs and GPUs, while maintaining matching accuracy and minimal footprint overhead.
- 1
- History-Aware and Dynamic Client Contribution in Federated Learning1
This work proposes a history-aware client contribution assessment framework, called FLContrib, where client-participation is dynamic, i.e., a subset of clients participates in each epoch, and applies FLContrib to detect dishonest clients in FL based on historical Shaplee values.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- On centripetal flows of entities in scale‐free networks with nodes of finite capability1
The results reinforce the importance of a network's hubs and their immediate environs, and suggest strategies for prioritizing elements of anetwork for optimization.
- 1
- 1
- 1
- An Integrated Approach to Speed Up GA-SVM Feature Selection Model1
An HPC-enabled GA-SVM (HGA-S VM) is proposed and implemented by integrating data parallelization, multithreading and heuristic techniques with the ultimate goal of maintaining robustness and lowering computational cost.
- Real-World Insights in Designing SteatoStat: An End-to-End Deep Learning Pipeline for Hepatic Steatosis Quantification–
These robust findings underscore the model’s potential clinical utility in providing a standardized objective quantification of hepatic steatosis and future directions include enhancing the model’s generalizability and its clinical integration through validation on independent, multi-institutional datasets.
- Confidence-Adaptive SwiGLU for Mixture-of-Experts–
Confidence-Aware SwiGLU ($\kappa$-SwiGLU), a variant of SwiGLU for Mixture-of-Experts (MoE) models that adjusts expert gate sharpness according to token-level routing confidence, is proposed, demonstrating that confidence-aware gate sharpness is a promising mechanism for improving MoE MLPs.
- Annotation-efficient medical image segmentation via cross-latent graphs and vector-quantized memory–
A framework for annotation-efficient medical segmentation that leverages sparse supervision from scribbles and points is proposed, which achieves competitive or superior performance compared with state-of-the-art weakly supervised methods, while approaching fully supervised accuracy.
- Performance and label efficiency of traditional deep-learning models and a retina-specific foundation model for ocular and systemic disease detection: a retrospective comparative study–
For systemic diseases, RETFound consistently outperformed traditional deep-learning models in internal testing when fine-tuned on smaller datasets and with smaller datasets, except in the case of diabetic retinopathy and glaucoma.
- Generative adversarial networks (GAN) for pre-dilation retinal photograph quality enhancement–
The enhanced images demonstrated improved quality and greater structural similarity to dilated images, suggesting the potential of the CofeNet model as an alternative approach to enhance image quality.
- –
- Structured Semantic Cloaking for Jailbreak Attacks on Large Language Models–
Structured Semantic Cloaking (S2C), a novel multi-dimensional jailbreak attack framework that manipulates how malicious semantic intent is reconstructed during model inference, is proposed and evaluated across multiple open-source and proprietary LLMs.
- –
- Aligning Medical Conversational AI through Online Reinforcement Learning with Information-Theoretic Rewards–
Information Gain Fine-Tuning (IGFT), a novel approach for training medical conversational AI to conduct effective patient interviews and generate comprehensive History of Present Illness (HPI) without requiring pre-collected human conversations, is presented.
- –
- –
- Enhancing Community Vision Screening: AI-Driven Retinal Photography for Early Disease Detection and Patient Trust–
The Enhancing Community Vision Screening (ECVS) solution is introduced, which addresses the aforementioned concerns with a novel and feasible solution based on simple, non-invasive retinal photography for the detection of pathology-based visual impairment.
- –
- –
- –
- Training-free image style alignment for self-adapting domain shift on handheld ultrasound devices–
The proposed TISA can directly infer handheld device images without extra training and is suited for clinical applications and shows that TISA performs better and more stably in medical detection and segmentation tasks for handheld device data.
- –
- RAPTOR: A Five-Safes approach to a secure, cloud native and serverless genomics data repository–
The Research Assets Provisioning and Tracking Online Repository (RAPTOR) by the Genome Institute of Singapore is a cloud native genomics data repository and analytics platform focusing on security and regulatory compliance.
- –
- –
- –
- –
- RAIN: Robust and Accurate Classification Networks with Randomization and Enhancement–
This paper proposes a novel framework to improve robustness and meanwhile retain the accuracy of given classification CNN models, termed as RAIN, which consists of two conjugate modules: structured randomization (SRd) and detail generation (DG).
- –
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar; position from the scholar’s ORCID record, last synced 2026-10-10. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it. You will be asked to sign in.