Previous works have shown that automatic speaker verification (ASV) is seriously vulnerable to malicious spoofing attacks, such as replay, synthetic speech, and recently emerged adversarial attacks. Great efforts have...
详细信息
Previous works have shown that automatic speaker verification (ASV) is seriously vulnerable to malicious spoofing attacks, such as replay, synthetic speech, and recently emerged adversarial attacks. Great efforts have been dedicated to defending ASV against replay and synthetic speech;however, only a few approaches have been explored to deal with adversarial attacks. All the existing approaches to tackle adversarial attacks for ASV require the knowledge for adversarial samples generation, but it is impractical for defenders to know the exact attack algorithms that are applied by the in-the-wild attackers. This work is among the first to perform adversarial defense for ASV without knowing the specific attack algorithms. Inspired by self-supervised learning models (SSLMs) that possess the merits of alleviating the superficial noise in the inputs and reconstructing clean samples from the interrupted ones, this work regards adversarial perturbations as one kind of noise and conducts adversarial defense for ASV by SSLMs. Specifically, we propose to perform adversarial defense from two perspectives: 1) adversarial perturbation purification and 2) adversarial perturbation detection. The purification module aims at alleviating the adversarial perturbations in the samples and pulling the contaminated adversarial inputs back towards the decision boundary. Experimental results show that our proposed purification module effectively counters adversarial attacks and outperforms traditional filters from both alleviating the adversarial noise and maintaining the performance of genuine samples. The detection module aims at detecting adversarial samples from genuine ones based on the statistical properties of ASV scores derived by a unique ASV integrating with different number of SSLMs. Experimental results show that our detection module helps shield the ASV by detecting adversarial samples. Both purification and detection methods are helpful for defending against different kinds of
作者：
Rabiha, Suciana GhadatiWibowo, AntoniLukasHeryadi, YayaComputer Science Department
BINUS Graduate Program-Doctor of Computer Science. Information Systems Department BINUS Online Learning Bina Nusantara University Jakarta11480 Indonesia Computer Science Department
BINUS Graduate Program-Doctor of Computer Science Bina Nusantara University 11480 Indonesia
Faculty of Engineering Universitas Katolik Indonesia Atma Jaya Indonesia Computer Science Department
BINUS Graduate Program - Doctor of Computer Science Bina Nusantara University 11480 Indonesia
One of the health problems that require special attention is diabetes, besides the growth of this disease infection is increasing in various circles ranging from children, adults, men, women and the elderly. So to det...
详细信息
Although using convolutional neural networks(CNNs)for computer-aided diagnosis(CAD)has made tremendous progress in the last few years,the small medical datasets remain to be the major bottleneck in this *** address th...
详细信息
Although using convolutional neural networks(CNNs)for computer-aided diagnosis(CAD)has made tremendous progress in the last few years,the small medical datasets remain to be the major bottleneck in this *** address this problem,researchers start looking for information out of the medical *** efforts mainly leverage information from natural images via transfer *** recent research work focuses on integrating knowledge from medical practitioners,either letting networks resemble how practitioners are trained,how they view images,or using extra *** this paper,we propose a scheme named Domain Guided-CNN(DG-CNN)to incorporate the margin information,a feature described in the consensus for radiologists to diagnose cancer in breast ultrasound(BUS)*** DG-CNN,attention maps that highlight margin areas of tumors are first generated,and then incorporated via different approaches into the *** have tested the performance of DG-CNN on our own dataset(including 1485 ultrasound images)and on a public *** results show that DG-CNN can be applied to different network structures like VGG and ResNet to improve their *** example,experimental results on our dataset show that with a certain integrating mode,the improvement of using DG-CNN over a baseline network structure ResNet 18 is 2.17%in accuracy,1.69%in sensitivity,2.64%in specificity and 2.57%in AUC(Area Under Curve).To the best of our knowledge,this is the first time that the margin information is utilized to improve the performance of deep neural networks in diagnosing breast cancer in BUS images.
As Metaverse emerges as the next-generation Internet paradigm, the ability to efficiently generate content is paramount. AI-Generated Content (AIGC) emerges as a key solution, yet the resource-intensive nature of larg...
详细信息
As Metaverse emerges as the next-generation Internet paradigm, the ability to efficiently generate content is paramount. AI-Generated Content (AIGC) emerges as a key solution, yet the resource-intensive nature of larg...
详细信息
We evaluate two schemes of service contracting (SC) program implemented for the first time in the Philippines: gross-cost and net-cost. Using an automatic vehicle location (AVL) dataset generated using a mobile applic...
详细信息
We evaluate two schemes of service contracting (SC) program implemented for the first time in the Philippines: gross-cost and net-cost. Using an automatic vehicle location (AVL) dataset generated using a mobile application, we examine the supply and performance of jeepney, a form of paratransit, on 20 routes during the pandemic. Ten of these routes are under gross-cost, and the other ten are under net-cost. Our evaluation suggests that gross-cost SC has resulted in more kilometer-run travelled, more arrivals at the stop, shorter headway duration, which are indicative of increased supply, compared to net-cost. However, we obtain mixed results for metrics associated with performance (headway regularity, dwell time and route compliance) for gross- and net- cost. We recommend that for SC to maximize its impact and raise performance, the government should set standards on metrics associated with fleet management and enforce strict compliance. Since SC is typically applied for the management of formal public transport, our work represents a novel contribution on the application of SC on the management of paratransit in the Global South.
As options for renewable procurement have proliferated to meet consumer demand, it is more complicated for consumers to navigate the available choices. In addition to installing distributed energy resources (e.g., sol...
详细信息
The demand for sustainable and renewable materials has prompted extensive research on lignocellulosic biomass as a potential feedstock for advanced nanomaterials. This article describes the sequential processes of iro...
详细信息
The demand for sustainable and renewable materials has prompted extensive research on lignocellulosic biomass as a potential feedstock for advanced nanomaterials. This article describes the sequential processes of iron-catalyzed graphitization, improved Hummer’s method, and low-temperature thermal annealing designed to transform ethanosolv lignin into bio-based graphene products. The lignin undergoes a metamorphosis during graphitization, transforming into turbostratic crystallite with a unique flake-like configuration at high temperatures in a tightly sealed muffle furnace. Following oxidation, the structure of the resultant graphite undergoes dynamic changes, leading to exfoliation-induced distancing of the sheets of graphene layers. Notwithstanding the reduction process, the geometrical morphology of its oxidized state remains unaltered, preserving the lamellar configuration. Notably, the resulting graphitized lignin showcases a substantial carbon content exceeding 75 wt % and a relatively low oxygen content below 16 wt %. In contrast, graphene oxides (GOs) exhibit elevated oxygen levels resulting from the introduction of oxygen moieties during the oxidation step, which subsequently diminishes following thermal annealing. Diffraction patterns confirm the presence of (001), (002), and (100) planes in all graphene-based materials, indicating an aromatic ring orientation. Oxidation affects interlayer spacing with GOs exhibiting larger distances than graphitized lignin and reduced graphene oxides (rGOs). Graphitization increases the sp2carbon composition by removing volatiles and mineral matter. Oxidation increases the sp3carbon composition, but it decreases after thermal annealing due to the partial removal of oxygen species, leading to a reorientation of the sp2carbon *** research explores converting bamboo lignin into customizable graphene materials for sustainable use. The study details processes of metamorphosis, oxidation, and reduction, reveali
Beta gallium oxide (b-Ga2O3) shows significant promise in the high-temperature, high-power, and sensing electronics applications. However, long-term stable metallization layers for Ohmic contacts at high temperature p...
详细信息
Currently, thirty five percent of high voltage transmission lines in Thailand are aged and degraded. These transmission lines need some renovation work with major components;i.e. conductor, insulator, steel structure,...
详细信息
暂无评论