Defensive Dual Masking for Robust Adversarial Defense

The field of textual adversarial defenses has gained considerable attention in recent years due to the increasing vulnerability of natural language processing (NLP) models to adversarial attacks, which exploit subtle perturbations in input text to deceive models. This paper introduces the Defensive Dual Masking (DDM) algorithm, a novel approach designed to enhance model robustness against such attacks. DDM utilizes a unique adversarial training strategy where [MASK] tokens are strategically inserted into training samples to prepare the model to handle adversarial perturbations more effectively. During inference, potentially adversarial tokens are dynamically replaced with [MASK] tokens to neutralize potential threats while preserving the core semantics of the input. The theoretical foundation of our approach is explored, demonstrating how the selective masking mechanism strengthens the model's ability to identify and mitigate adversarial manipulations. Our empirical evaluation across a diverse set of benchmark datasets and attack mechanisms consistently shows that DDM outperforms state-of-the-art defense techniques, improving model accuracy and robustness. Moreover, when applied to Large Language Models (LLMs), DDM also enhances their resilience to adversarial attacks, providing a scalable defense mechanism for large-scale NLP applications.

Vox-UDA: Voxel-wise Unsupervised Domain Adaptation for Cryo-Electron Subtomogram Segmentation with Denoised Pseudo Labeling

Cryo-Electron Tomography (cryo-ET) is a 3D imaging technology facilitating the study of macromolecular structures at near-atomic resolution. Recent volumetric segmentation approaches on cryo-ET images have drawn widespread interest in biological sector. However, existing methods heavily rely on manually labeled data, which requires highly professional skills, thereby hindering the adoption of fully-supervised approaches for cryo-ET images. Some unsupervised domain adaptation (UDA) approaches have been designed to enhance the segmentation network performance using unlabeled data. However, applying these methods directly to cryo-ET images segmentation tasks remains challenging due to two main issues: 1) the source data, usually obtained through simulation, contain a certain level of noise, while the target data, directly collected from raw-data from real-world scenario, have unpredictable noise levels. 2) the source data used for training typically consists of known macromoleculars, while the target domain data are often unknown, causing the model's segmenter to be biased towards these known macromolecules, leading to a domain shift problem. To address these challenges, in this work, we introduce the first voxel-wise unsupervised domain adaptation approach, termed Vox-UDA, specifically for cryo-ET subtomogram segmentation. Vox-UDA incorporates a noise generation module to simulate target-like noises in the source dataset for cross-noise level adaptation. Additionally, we propose a denoised pseudo-labeling strategy based on improved Bilateral Filter to alleviate the domain shift problem. Experimental results on both simulated and real cryo-ET subtomogram datasets demonstrate the superiority of our proposed approach compared to state-of-the-art UDA methods.

Regression on Deep Visual Features using Artificial Neural Networks (ANNs) to Predict Hydraulic Blockage at Culverts

Cross drainage hydraulic structures (i.e., culverts, bridges) in urban landscapes are prone to getting blocked by transported debris which often results in causing the flash floods. In context of Australia, Wollongong City Council (WCC) blockage conduit policy is the only formal guideline to consider blockage in design process. However, many argue that this policy is based on the post floods visual inspections and hence can not be considered accurate representation of hydraulic blockage. As a result of this on-going debate, visual blockage and hydraulic blockage are considered two distinct terms with no established quantifiable relation among both. This paper attempts to relate both terms by proposing the use of deep visual features for prediction of hydraulic blockage at a given culvert. An end-to-end machine learning pipeline is propounded which takes an image of culvert as input, extract visual features using deep learning models, pre-process the visual features and feed into regression model to predict the corresponding hydraulic blockage. Dataset (i.e., Hydrology-Lab Dataset (HD), Visual Hydrology-Lab Dataset (VHD)) used in this research was collected from in-lab experiments carried out using scaled physical models of culverts where multiple blockage scenarios were replicated at scale. Performance of regression models was assessed using standard evaluation metrics. Furthermore, performance of overall machine learning pipeline was assessed in terms of processing times for relative comparison of models and hardware requirement analysis. From the results ANN used with MobileNet extracted visual features achieved the best regression performance with $R^{2}$ score of 0.7855. Positive value of $R^{2}$ score indicated the presence of correlation between visual features and hydraulic blockage and suggested that both can be interrelated with each other.

Automating Visual Blockage Classification of Culverts with Deep Learning

Blockage of culverts by transported debris materials is reported as main contributor in originating urban flash floods. Conventional modelling approaches had no success in addressing the problem largely because of unavailability of peak floods hydraulic data and highly non-linear behaviour of debris at culvert. This article explores a new dimension to investigate the issue by proposing the use of Intelligent Video Analytic (IVA) algorithms for extracting blockage related information. Potential of using existing Convolutional Neural Network (CNN) algorithms (i.e., DarkNet53, DenseNet121, InceptionResNetV2, InceptionV3, MobileNet, ResNet50, VGG16, EfficientNetB3, NASNet) is investigated over a custom collected blockage dataset (i.e., Images of Culvert Openings and Blockage (ICOB)) to predict the blockage in a given image. Models were evaluated based on their performance on test dataset (i.e., accuracy, loss, precision, recall, F1-score, Jaccard-Index), Floating Point Operations Per Second (FLOPs) and response times to process a single test instance. From the results, NASNet was reported most efficient in classifying the blockage with the accuracy of 85\%; however, EfficientNetB3 was recommended for the hardware implementation because of its improved response time with accuracy comparable to NASNet (i.e., 83\%). False Negative (FN) instances, False Positive (FP) instances and CNN layers activation suggested that background noise and oversimplified labelling criteria were two contributing factors in degraded performance of existing CNN algorithms.

Prediction of Hydraulic Blockage at Cross Drainage Structures using Regression Analysis

Hydraulic blockage of cross-drainage structures such as culverts is considered one of main contributor in triggering urban flash floods. However, due to lack of during floods data and highly non-linear nature of debris interaction, conventional modelling for hydraulic blockage is not possible. This paper proposes to use machine learning regression analysis for the prediction of hydraulic blockage. Relevant data has been collected by performing a scaled in-lab study and replicating different blockage scenarios. From the regression analysis, Artificial Neural Network (ANN) was reported best in hydraulic blockage prediction with $R^2$ of 0.89. With deployment of hydraulic sensors in smart cities, and availability of Big Data, regression analysis may prove helpful in addressing the blockage detection problem which is difficult to counter using conventional experimental and hydrological approaches.