Bharadwaj Veeravalli

Soumyabrata Dev, Hewei Wang, Chidozie Shamrock Nwosu et al.

The negative impact of stroke in society has led to concerted efforts to improve the management and diagnosis of stroke. With an increased synergy between technology and medical diagnosis, caregivers create opportunities for better patient management by systematically mining and archiving the patients' medical records. Therefore, it is vital to study the interdependency of these risk factors in patients' health records and understand their relative contribution to stroke prediction. This paper systematically analyzes the various factors in electronic health records for effective stroke prediction. Using various statistical techniques and principal component analysis, we identify the most important factors for stroke prediction. We conclude that age, heart disease, average glucose level, and hypertension are the most important factors for detecting stroke in patients. Furthermore, a perceptron neural network using these four attributes provides the highest accuracy rate and lowest miss rate compared to using all available input features and other benchmarking algorithms. As the dataset is highly imbalanced concerning the occurrence of stroke, we report our results on a balanced dataset created via sub-sampling techniques.

Abhishek Tyagi, Yunuo Cen, Shrey Dhorajiya et al.

Large Language Models (LLMs) exhibit substantial parameter redundancy, particularly in Feed-Forward Networks (FFNs). Existing pruning methods suffer from two primary limitations. First, reliance on dataset-specific calibration introduces significant data dependency and computational overhead. Second, being predominantly static, they fail to account for the evolving subset of knowledge neurons in LLMs during autoregressive generation as the context evolves. To address this, we introduce DART, i.e., Dynamic Attention-Guided Runtime Tracing), a lightweight, training-free method that performs on-the-fly context-based pruning. DART monitors shifts in attention score distributions to infer context changes, dynamically updating neuron-level masks to retain salient parameters. Across ten benchmarks, DART outperforms prior dynamic baseline, achieving accuracy gains of up to 14.5% on LLAMA-3.1-8B at 70% FFN sparsity. Furthermore, DART achieves up to 3x better ROUGE-L scores with respect to static-masked pruning on summarization tasks, with its performance comparable to the original dense models. We conclusively demonstrate that the proposed framework effectively adapts to diverse semantic contexts, preserves model capabilities across both general and domain-specific tasks while running at less than 10MBs of memory for LLAMA-3.1-8B(16GBs) with 0.1% FLOPs overhead. The code is available at https://github.com/seeder-research/DART.

Mikhail Kennerley, Jian-Gang Wang, Bharadwaj Veeravalli et al.

Object detection at night is a challenging problem due to the absence of night image annotations. Despite several domain adaptation methods, achieving high-precision results remains an issue. False-positive error propagation is still observed in methods using the well-established student-teacher framework, particularly for small-scale and low-light objects. This paper proposes a two-phase consistency unsupervised domain adaptation network, 2PCNet, to address these issues. The network employs high-confidence bounding-box predictions from the teacher in the first phase and appends them to the student's region proposals for the teacher to re-evaluate in the second phase, resulting in a combination of high and low confidence pseudo-labels. The night images and pseudo-labels are scaled-down before being used as input to the student, providing stronger small-scale pseudo-labels. To address errors that arise from low-light regions and other night-related attributes in images, we propose a night-specific augmentation pipeline called NightAug. This pipeline involves applying random augmentations, such as glare, blur, and noise, to daytime images. Experiments on publicly available datasets demonstrate that our method achieves superior results to state-of-the-art methods by 20\%, and to supervised models trained directly on the target data.

Ziyuan Zhao, Jinxuan Hu, Zeng Zeng et al.

With large-scale well-labeled datasets, deep learning has shown significant success in medical image segmentation. However, it is challenging to acquire abundant annotations in clinical practice due to extensive expertise requirements and costly labeling efforts. Recently, contrastive learning has shown a strong capacity for visual representation learning on unlabeled data, achieving impressive performance rivaling supervised learning in many domains. In this work, we propose a novel multi-scale multi-view global-local contrastive learning (MMGL) framework to thoroughly explore global and local features from different scales and views for robust contrastive learning performance, thereby improving segmentation performance with limited annotations. Extensive experiments on the MM-WHS dataset demonstrate the effectiveness of MMGL framework on semi-supervised cardiac image segmentation, outperforming the state-of-the-art contrastive learning methods by a large margin.

Ziyuan Zhao, Wenjing Lu, Zeng Zeng et al.

Label scarcity has been a long-standing issue for biomedical image segmentation, due to high annotation costs and professional requirements. Recently, active learning (AL) strategies strive to reduce annotation costs by querying a small portion of data for annotation, receiving much traction in the field of medical imaging. However, most of the existing AL methods have to initialize models with some randomly selected samples followed by active selection based on various criteria, such as uncertainty and diversity. Such random-start initialization methods inevitably introduce under-value redundant samples and unnecessary annotation costs. For the purpose of addressing the issue, we propose a novel self-supervised assisted active learning framework in the cold-start setting, in which the segmentation model is first warmed up with self-supervised learning (SSL), and then SSL features are used for sample selection via latent feature clustering without accessing labels. We assess our proposed methodology on skin lesions segmentation task. Extensive experiments demonstrate that our approach is capable of achieving promising performance with substantial improvements over existing baselines.

Ziyuan Zhao, Andong Zhu, Zeng Zeng et al.

While deep models have shown promising performance in medical image segmentation, they heavily rely on a large amount of well-annotated data, which is difficult to access, especially in clinical practice. On the other hand, high-accuracy deep models usually come in large model sizes, limiting their employment in real scenarios. In this work, we propose a novel asymmetric co-teacher framework, ACT-Net, to alleviate the burden on both expensive annotations and computational costs for semi-supervised knowledge distillation. We advance teacher-student learning with a co-teacher network to facilitate asymmetric knowledge distillation from large models to small ones by alternating student and teacher roles, obtaining tiny but accurate models for clinical employment. To verify the effectiveness of our ACT-Net, we employ the ACDC dataset for cardiac substructure segmentation in our experiments. Extensive experimental results demonstrate that ACT-Net outperforms other knowledge distillation methods and achieves lossless segmentation performance with 250x fewer parameters.

Xu Kaixin, Liu Liyang, Zhao Ziyuan et al.

Multi-label Learning on Image data has been widely exploited with deep learning models. However, supervised training on deep CNN models often cannot discover sufficient discriminative features for classification. As a result, numerous self-supervision methods are proposed to learn more robust image representations. However, most self-supervised approaches focus on single-instance single-label data and fall short on more complex images with multiple objects. Therefore, we propose an Object-Aware Self-Supervision (OASS) method to obtain more fine-grained representations for multi-label learning, dynamically generating auxiliary tasks based on object locations. Secondly, the robust representation learned by OASS can be leveraged to efficiently generate Class-Specific Instances (CSI) in a proposal-free fashion to better guide multi-label supervision signal transfer to instances. Extensive experiments on the VOC2012 dataset for multi-label classification demonstrate the effectiveness of the proposed method against the state-of-the-art counterparts.

Bharadwaj Veeravalli

We develop an integrated Multi-Port Concurrent Communication Divisible Load Theory (MPCC-DLT) framework for relay-centric distributed satellite systems (DSS), capturing concurrent data dissemination, parallel computation, and result return under heterogeneous onboard processing and inter-satellite link conditions. We propose a formulation that yields closed-form expressions for optimal load allocation and completion time that explicitly quantify the joint impact of computation speed, link bandwidth, and result-size overhead. We further derive deadline feasibility conditions that enable explicit sizing of cooperative satellite clusters to meet time-critical task requirements. Extensive simulation results demonstrate that highly distributable tasks achieve substantial latency reduction, while communication-heavy tasks exhibit diminishing returns due to result-transfer overheads. To bridge theory and practice, we extend the MPCC-DLT framework with a real-time admission control mechanism that handles stochastic task arrivals and deadline constraints, enabling blocking-aware operation. Our real-time simulations illustrate how task structure and system parameters jointly govern deadline satisfaction and operating regimes. Overall, this work provides the first analytically tractable MPCC-DLT model for distributed satellite systems and offers actionable insights for application-aware scheduling and system-level design of future satellite constellations.

Bharadwaj Veeravalli

We present the design of a Resource-Aware Task Allocator (RATA) and an empirical analysis in handling real-time tasks for processing on Distributed Satellite Systems (DSS). We consider task processing performance across low Earth orbit (LEO) to Low-Medium Earth Orbit (Low-MEO) constellation sizes, under varying traffic loads. Using Single-Level Tree Network(SLTN)-based cooperative task allocation architecture, we attempt to evaluate some key performance metrics - blocking probabilities, response times, energy consumption, and resource utilization across several tens of thousands of tasks per experiment. Our resource-conscious RATA monitors key parameters such as arrival rate, resources (on-board compute, storage, bandwidth, battery) availability, satellite eclipses' influence in processing and communications. This study is an important step towards analyzing the performance under lighter to stress inducing levels of compute intense workloads to test the ultimate performance limits under the combined influence of the above-mentioned factors. Results show pronounced non-linear scaling: while capacity increases with constellation size, blocking and delay grow rapidly, whereas energy remains resilient under solar-aware scheduling. The analysis identifies a practical satellite-count limit for baseline SLTNs and demonstrates that CPU availability, rather than energy, is the primary cause of blocking. These findings provide quantitative guidance by identifying thresholds at which system performance shifts from graceful degradation to collapse.

Bharadwaj Veeravalli

In this paper, we introduce the first machine learning framework for predicting optimal processing times in Single-Level Tree Network (SLTN) architectures for the Divisible Load Theory (DLT) paradigm. Using a feedforward neural network(FNN) with 16 engineered features, we train a model on 100,000 synthetically generated configurations to predict optimal processing times without explicit formulation of DLT equations. The model achieves 97-99% accuracy (R-square factor) with mean absolute percentage error of 1-5%, demonstrating that neural networks can effectively learn complex load distribution relationships. Feature importance analysis reveals that the model implicitly captures DLT mathematical structure, including load conservation and simultaneous finishing constraints. With inference times under 1 millisecond, the approach provides 10-100x speedup over traditional DLT computation, enabling applications in real-time scheduling, design space exploration, and cloud resource allocation. The method generalizes well across diverse system configurations (n=3 to 20, load size =1 to 100 GB) with consistent accuracy, though performance degrades slightly for very large or highly heterogeneous systems. This work demonstrates the feasibility of using machine learning to accelerate distributed computing optimization while maintaining near-optimal accuracy.

Qilong Wu, Xiaoneng Xiang, Hejia Huang et al.

The rapid growth of the financial sector and the rising focus on Environmental, Social, and Governance (ESG) considerations highlight the need for advanced NLP tools. However, open-source LLMs proficient in both finance and ESG domains remain scarce. To address this gap, we introduce SusGen-30K, a category-balanced dataset comprising seven financial NLP tasks and ESG report generation, and propose TCFD-Bench, a benchmark for evaluating sustainability report generation. Leveraging this dataset, we developed SusGen-GPT, a suite of models achieving state-of-the-art performance across six adapted and two off-the-shelf tasks, trailing GPT-4 by only 2% despite using 7-8B parameters compared to GPT-4's 1,700B. Based on this, we propose the SusGen system, integrated with Retrieval-Augmented Generation (RAG), to assist in sustainability report generation. This work demonstrates the efficiency of our approach, advancing research in finance and ESG.

Jingze Ge, Xue Geng, Yun Liu et al.

To mitigate the memory constraints associated with fine-tuning large pre-trained models, existing parameter-efficient fine-tuning (PEFT) methods, such as LoRA, rely on low-rank updates. However, such updates fail to fully capture the rank characteristics of the weight modifications observed in full-parameter fine-tuning, resulting in a performance gap. Furthermore, LoRA and other existing PEFT methods still require substantial memory to store the full set of frozen weights, limiting their efficiency in resource-constrained settings. To addres these limitations, we introduce Cumulative Energy-Retaining Subspace Adaptation (CERSA), a novel fine-tuning paradigm that leverages singular value decomposition (SVD) to retain only the principal components responsible for 90% to 95% of the spectral energy. By fine-tuning low-rank representations derived from this principal subspace, CERSA significantly reduces memory consumption. We conduct extensive evaluations of CERSA across models of varying scales and domains, including image recognition, text-to-image generation, and natural language understanding. Empirical results demonstrate that CERSA consistently outperforms state-of-the-art PEFT methods while achieving substantially lower memory requirements. The code will be publicly released.

Mikhail Kennerley, Jian-Gang Wang, Bharadwaj Veeravalli et al.

Domain adaptive object detection aims to adapt detection models to domains where annotated data is unavailable. Existing methods have been proposed to address the domain gap using the semi-supervised student-teacher framework. However, a fundamental issue arises from the class imbalance in the labelled training set, which can result in inaccurate pseudo-labels. The relationship between classes, especially where one class is a majority and the other minority, has a large impact on class bias. We propose Class-Aware Teacher (CAT) to address the class bias issue in the domain adaptation setting. In our work, we approximate the class relationships with our Inter-Class Relation module (ICRm) and exploit it to reduce the bias within the model. In this way, we are able to apply augmentations to highly related classes, both inter- and intra-domain, to boost the performance of minority classes while having minimal impact on majority classes. We further reduce the bias by implementing a class-relation weight to our classification loss. Experiments conducted on various datasets and ablation studies show that our method is able to address the class bias in the domain adaptation setting. On the Cityscapes to Foggy Cityscapes dataset, we attained a 52.5 mAP, a substantial improvement over the 51.2 mAP achieved by the state-of-the-art method.

Zhikan Wang, Zhongyao Cheng, Jiajie Xiong et al.

In recent years, the rapid advancement of deepfake technology has revolutionized content creation, lowering forgery costs while elevating quality. However, this progress brings forth pressing concerns such as infringements on individual rights, national security threats, and risks to public safety. To counter these challenges, various detection methodologies have emerged, with Vision Transformer (ViT)-based approaches showcasing superior performance in generality and efficiency. This survey presents a timely overview of ViT-based deepfake detection models, categorized into standalone, sequential, and parallel architectures. Furthermore, it succinctly delineates the structure and characteristics of each model. By analyzing existing research and addressing future directions, this survey aims to equip researchers with a nuanced understanding of ViT's pivotal role in deepfake detection, serving as a valuable reference for both academic and practical pursuits in this domain.

Hugo Chan-To-Hing, Bharadwaj Veeravalli

Self-supervised frameworks for representation learning have recently stirred up interest among the remote sensing community, given their potential to mitigate the high labeling costs associated with curating large satellite image datasets. In the realm of multimodal data fusion, while the often used contrastive learning methods can help bridging the domain gap between different sensor types, they rely on data augmentations techniques that require expertise and careful design, especially for multispectral remote sensing data. A possible but rather scarcely studied way to circumvent these limitations is to use a masked image modelling based pretraining strategy. In this paper, we introduce Fus-MAE, a self-supervised learning framework based on masked autoencoders that uses cross-attention to perform early and feature-level data fusion between synthetic aperture radar and multispectral optical data - two modalities with a significant domain gap. Our empirical findings demonstrate that Fus-MAE can effectively compete with contrastive learning strategies tailored for SAR-optical data fusion and outperforms other masked-autoencoders frameworks trained on a larger corpus.

Siwen Jiao, Yangyi Fang, Baoyun Peng et al.

Recent advancements in Visual Language Models (VLMs) have made them crucial for visual question answering (VQA) in autonomous driving, enabling natural human-vehicle interactions. However, existing methods often struggle in dynamic driving environments, as they usually focus on static images or videos and rely on downsampling to manage computational costs. This results in the loss of critical details and the difficulty in effectively integrating spatial and temporal information, undermining fine-grained perception and temporal coherence essential for effective decision-making. To tackle these challenges, we introduce LaVida Drive, a novel and efficient VQA framework for autonomous driving. LaVida Drive seamlessly integrates temporal data while maintaining high-resolution inputs for detailed visual perception. It optimizes spatial processing by retaining high-resolution data for intricate details and using lower-resolution inputs for temporal analysis to focus on motion-related features, thereby boosting computational efficiency. The core of LaVida Drive consists of two modules: the \textit{Query-aware Token Selection} module and the \textit{Spatial-Temporal Token Recovery and Enhancement} module. The former dynamically selects the most relevant visual tokens based on semantic alignment with the input query, reducing the token count from high-resolution spatial input. The latter ensures smooth and coherent interactions between spatial and temporal information, preserving contextual continuity across frames. Extensive experiments on various autonomous driving question-answering benchmarks show that LaVida Drive significantly reduces visual tokens, enhances efficiency, and improves overall performance.

Lechun You, Zhonghua Wu, Weide Liu et al.

Current methods for 3D semantic segmentation propose training models with limited annotations to address the difficulty of annotating large, irregular, and unordered 3D point cloud data. They usually focus on the 3D domain only, without leveraging the complementary nature of 2D and 3D data. Besides, some methods extend original labels or generate pseudo labels to guide the training, but they often fail to fully use these labels or address the noise within them. Meanwhile, the emergence of comprehensive and adaptable foundation models has offered effective solutions for segmenting 2D data. Leveraging this advancement, we present a novel approach that maximizes the utility of sparsely available 3D annotations by incorporating segmentation masks generated by 2D foundation models. We further propagate the 2D segmentation masks into the 3D space by establishing geometric correspondences between 3D scenes and 2D views. We extend the highly sparse annotations to encompass the areas delineated by 3D masks, thereby substantially augmenting the pool of available labels. Furthermore, we apply confidence- and uncertainty-based consistency regularization on augmentations of the 3D point cloud and select the reliable pseudo labels, which are further spread on the 3D masks to generate more labels. This innovative strategy bridges the gap between limited 3D annotations and the powerful capabilities of 2D foundation models, ultimately improving the performance of 3D weakly supervised segmentation.

Thao N. N. Nguyen, Bharadwaj Veeravalli, Xuanyao Fong

Long training time hinders the potential of the deep, large-scale Spiking Neural Network (SNN) with the on-chip learning capability to be realized on the embedded systems hardware. Our work proposes a novel connection pruning approach that can be applied during the on-chip Spike Timing Dependent Plasticity (STDP)-based learning to optimize the learning time and the network connectivity of the deep SNN. We applied our approach to a deep SNN with the Time To First Spike (TTFS) coding and has successfully achieved 2.1x speed-up and 64% energy savings in the on-chip learning and reduced the network connectivity by 92.83%, without incurring any accuracy loss. Moreover, the connectivity reduction results in 2.83x speed-up and 78.24% energy savings in the inference. Evaluation of our proposed approach on the Field Programmable Gate Array (FPGA) platform revealed 0.56% power overhead was needed to implement the pruning algorithm.

Jiapan Gu, Ziyuan Zhao, Zeng Zeng et al.

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer and the fourth most common cause of cancer-related death worldwide. Understanding the underlying gene mutations in HCC provides great prognostic value for treatment planning and targeted therapy. Radiogenomics has revealed an association between non-invasive imaging features and molecular genomics. However, imaging feature identification is laborious and error-prone. In this paper, we propose an end-to-end deep learning framework for mutation prediction in APOB, COL11A1 and ATRX genes using multiphasic CT scans. Considering intra-tumour heterogeneity (ITH) in HCC, multi-region sampling technology is implemented to generate the dataset for experiments. Experimental results demonstrate the effectiveness of the proposed model.

Ziyuan Zhao, Xiaoyan Yang, Bharadwaj Veeravalli et al.

Segmentation is a prerequisite yet challenging task for medical image analysis. In this paper, we introduce a novel deeply supervised active learning approach for finger bones segmentation. The proposed architecture is fine-tuned in an iterative and incremental learning manner. In each step, the deep supervision mechanism guides the learning process of hidden layers and selects samples to be labeled. Extensive experiments demonstrated that our method achieves competitive segmentation results using less labeled samples as compared with full annotation.