Maisha Binte Rashid

Maisha Binte Rashid, Pablo Rivas

Multimodal machine learning models that combine visual and textual data are increasingly being deployed in critical applications, raising significant safety and security concerns due to their vulnerability to adversarial attacks. This paper presents an effective strategy to enhance the robustness of multimodal image captioning models against such attacks. By leveraging the Fast Gradient Sign Method (FGSM) to generate adversarial examples and incorporating adversarial training techniques, we demonstrate improved model robustness on two benchmark datasets: Flickr8k and COCO. Our findings indicate that selectively training only the text decoder of the multimodal architecture shows performance comparable to full adversarial training while offering increased computational efficiency. This targeted approach suggests a balance between robustness and training costs, facilitating the ethical deployment of multimodal AI systems across various domains.

Maisha Binte Rashid, Pablo Rivas

Vision-Language Models (VLMs) are increasingly deployed in public sector missions, necessitating robust evaluation of their safety and vulnerability to adversarial attacks. This paper introduces a novel framework to quantify adversarial risks in VLMs. We analyze model performance under Gaussian, salt-and-pepper, and uniform noise, identifying misclassification thresholds and deriving composite noise patches and saliency patterns that highlight vulnerable regions. These patterns are compared against the Fast Gradient Sign Method (FGSM) to assess their adversarial effectiveness. We propose a new Vulnerability Score that combines the impact of random noise and adversarial attacks, providing a comprehensive metric for evaluating model robustness.

Maisha Binte Rashid, Pablo Rivas

In this paper, we aim to investigate the capabilities of multimodal machine learning models, particularly the OpenFlamingo model, in processing a large-scale dataset of consumer-to-consumer (C2C) online posts related to car parts. We have collected data from two platforms, OfferUp and Craigslist, resulting in a dataset of over 1.2 million posts with their corresponding images. The OpenFlamingo model was used to extract embeddings for the text and image of each post. We used $k$-means clustering on the joint embeddings to identify underlying patterns and commonalities among the posts. We have found that most clusters contain a pattern, but some clusters showed no internal patterns. The results provide insight into the fact that OpenFlamingo can be used for finding patterns in large datasets but needs some modification in the architecture according to the dataset.