Camila Roa
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Camila Roa has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Michela Taufer
University of Tennessee, Knoxville, Knoxville, TN, USA
2 shared publications - Paula Olaya
The University of Tennessee, Knoxville, US
2 shared publications - Ricardo M. Llamas
University of Delaware
2 shared publications - Rodrigo Vargas
Department of Plant and Soil Sciences University of Delaware Newark Delaware USA
2 shared publications - Alana Romanella
OIT, University of Colorado, Boulder, CO, USA
1 shared publication - Amir Sadovnik
Oak Ridge National Laboratory, Oak Ridge, Tennessee, USA
1 shared publication
Latest Publications
- Open Access
- 6 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 42-52
Semantic Stealth: Crafting Covert Adversarial Patches for Sentiment Classifiers Using Large Language Models
Deep learning models have been shown to be vulnerable to adversarial attacks, in which perturbations to their inputs cause the model to produce incorrect predictions. As opposed to adversarial attacks in computer vision, where small changes introduced to pixel values can drastically alter a model's output while remaining imperceptible to humans, text-based attacks are difficult to conceal due to the discrete nature of tokens. Consequently, unconstrained gradient-based attacks often produce adversarial examples that lack semantic meaning, rendering them detectable through visual inspection or perplexity filters. In contrast to methods that rely on gradient-based optimization in the embedding space, we propose an approach that leverages a Large Language Model's ability to generate grammatically correct and semantically meaningful text to craft adversarial patches that seamlessly blend in with the original input text. These patches can be used to alter the behavior of a target model, such as a text classifier. Since our approach does not rely on gradient backpropagation, it only requires access to the target model's confidence scores, making it a grey-box attack. We demonstrate the feasibility of our approach using open-source LLMs, including Intel's Neural Chat, Llama2, and Mistral-Instruct, to generate adversarial patches capable of altering the predictions of a distilBERT model fine-tuned on the IMDB reviews dataset for sentiment classification.
- 6 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 42-52
Open AccessSemantic Stealth: Crafting Covert Adversarial Patches for Sentiment Classifiers Using Large Language Models
Deep learning models have been shown to be vulnerable to adversarial attacks, in which perturbations to their inputs cause the model to produce incorrect predictions. As opposed to adversarial attacks in computer vision, where small changes introduced to pixel values can drastically alter a model's output while remaining imperceptible to humans, text-based attacks are difficult to conceal due to the discrete nature of tokens. Consequently, unconstrained gradient-based attacks often produce adversarial examples that lack semantic meaning, rendering them detectable through visual inspection or perplexity filters. In contrast to methods that rely on gradient-based optimization in the embedding space, we propose an approach that leverages a Large Language Model's ability to generate grammatically correct and semantically meaningful text to craft adversarial patches that seamlessly blend in with the original input text. These patches can be used to alter the behavior of a target model, such as a text classifier. Since our approach does not rely on gradient backpropagation, it only requires access to the target model's confidence scores, making it a grey-box attack. We demonstrate the feasibility of our approach using open-source LLMs, including Intel's Neural Chat, Llama2, and Mistral-Instruct, to generate adversarial patches capable of altering the predictions of a distilBERT model fine-tuned on the IMDB reviews dataset for sentiment classification.
- Open Access
- 4 December 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-9
End-to-end Integration of Scientific Workflows on Distributed Cyberinfrastructures: Challenges and Lessons Learned with an Earth Science Application
Distributed cyberinfrastructures (CI) pose opportunities and challenges for the execution of scientific workflows, especially in the context of Earth science applications. They provide heterogeneous resources that can meet the needs of the applications that are part of the scientific workflows and provide the necessary performance and scalability to achieve scientific goals. However, the challenge with distributed CI is that it is difficult to find the right resources for the applications and to orchestrate the workflow execution from resource provisioning to job execution to delivering the final results. In some cases, poor choice of resources may result in slow execution or outright failure. In this paper, we present Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) Pegasus, a CI solution built as part of the U.S. National Science Foundation ACCESS program that provides automated execution of scientific applications. We demonstrate Pegasus's capabilities with SOil MOisture SPatial Inference Engine (SOMOSPIE), an earth science multi-component application for fine-grained soil moisture predictions. We identify a roadmap to migrate applications such as SOMOSPIE on ACCESS resources with the support of ACCESS Pegasus, outlining both strengths and weaknesses of this approach.
- 4 December 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Open AccessEnd-to-end Integration of Scientific Workflows on Distributed Cyberinfrastructures: Challenges and Lessons Learned with an Earth Science Application
Distributed cyberinfrastructures (CI) pose opportunities and challenges for the execution of scientific workflows, especially in the context of Earth science applications. They provide heterogeneous resources that can meet the needs of the applications that are part of the scientific workflows and provide the necessary performance and scalability to achieve scientific goals. However, the challenge with distributed CI is that it is difficult to find the right resources for the applications and to orchestrate the workflow execution from resource provisioning to job execution to delivering the final results. In some cases, poor choice of resources may result in slow execution or outright failure. In this paper, we present Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) Pegasus, a CI solution built as part of the U.S. National Science Foundation ACCESS program that provides automated execution of scientific applications. We demonstrate Pegasus's capabilities with SOil MOisture SPatial Inference Engine (SOMOSPIE), an earth science multi-component application for fine-grained soil moisture predictions. We identify a roadmap to migrate applications such as SOMOSPIE on ACCESS resources with the support of ACCESS Pegasus, outlining both strengths and weaknesses of this approach.
- Open Access
- 7 August 2023
- Published by Association for Computing Machinery (ACM)
- p. 311-312
GEOtiled: A Scalable Workflow for Generating Large Datasets of High-Resolution Terrain Parameters
Terrain parameters such as slope, aspect, and hillshading are essential in various applications, including agriculture, forestry, and hydrology. However, generating high-resolution terrain parameters is computationally intensive, making it challenging to provide these value-added products to communities in need. We present a scalable workflow called GEOtiled that leverages data partitioning to accelerate the computation of terrain parameters from digital elevation models, while preserving accuracy. We assess our workflow in terms of its accuracy and wall time by comparing it to SAGA, which is highly accurate but slow to generate results, and to GDAL, which supports memory optimizations but not data parallelism. We obtain a coefficient of determination (R^2) between GEOtiled and SAGA of 0.794, ensuring accuracy in our terrain parameters. We achieve an X6 speedup compared to GDAL when generating the terrain parameters at a high-resolution (10 m) for the Contiguous United States (CONUS).
- 7 August 2023
- Published by Association for Computing Machinery (ACM)
- p. 311-312
Open AccessGEOtiled: A Scalable Workflow for Generating Large Datasets of High-Resolution Terrain Parameters
Terrain parameters such as slope, aspect, and hillshading are essential in various applications, including agriculture, forestry, and hydrology. However, generating high-resolution terrain parameters is computationally intensive, making it challenging to provide these value-added products to communities in need. We present a scalable workflow called GEOtiled that leverages data partitioning to accelerate the computation of terrain parameters from digital elevation models, while preserving accuracy. We assess our workflow in terms of its accuracy and wall time by comparing it to SAGA, which is highly accurate but slow to generate results, and to GDAL, which supports memory optimizations but not data parallelism. We obtain a coefficient of determination (R^2) between GEOtiled and SAGA of 0.794, ensuring accuracy in our terrain parameters. We achieve an X6 speedup compared to GDAL when generating the terrain parameters at a high-resolution (10 m) for the Contiguous United States (CONUS).