publications

2024

1. Learning To See But Forgetting To Follow: Visual Instruction Tuning Makes LLMs More Prone To Jailbreak Attacks

   Georgios Pantazopoulos, Amit Parekh, Malvina Nikandrou, and 1 more author

   In Proceedings of Safety4ConvAI: The Third Workshop on Safety for Conversational AI @ LREC-COLING 2024, May 2024

   Abs

   Augmenting Large Language Models (LLMs) with image-understanding capabilities has resulted in a boom of high-performing Vision-Language models (VLMs). While studying the alignment of LLMs to human values has received widespread attention, the safety of VLMs has not received the same attention. In this paper, we explore the impact of jailbreaking on three state-of-the-art VLMs, each using a distinct modeling approach. By comparing each VLM to their respective LLM backbone, we find that each VLM is more susceptible to jailbreaking. We consider this as an undesirable outcome from visual instruction-tuning, which imposes a forgetting effect on an LLM’s safety guardrails. Therefore, we provide recommendations for future work based on evaluation strategies that aim to highlight the weaknesses of a VLM, as well as take safety measures into account during visual instruction tuning.

2. Enhancing Continual Learning in Visual Question Answering with Modality-Aware Feature Distillation

   Malvina Nikandrou, Georgios Pantazopoulos, Ioannis Konstas, and 1 more author

   In Proceedings of the 3rd Workshop on Advances in Language and Vision Research (ALVR), Aug 2024

   Abs

   Continual learning focuses on incrementally training a model on a sequence of tasks with the aim of learning new tasks while minimizing performance drop on previous tasks. Existing approaches at the intersection of Continual Learning and Visual Question Answering (VQA) do not study how the multimodal nature of the input affects the learning dynamics of a model. In this paper, we demonstrate that each modality evolves at different rates across a continuum of tasks and that this behavior occurs in established encoder-only models as well as modern recipes for developing Vision & Language (VL) models. Motivated by this observation, we propose a modality-aware feature distillation (MAFED) approach which outperforms existing baselines across models of varying scale in three multimodal continual learning settings. Furthermore, we provide ablations showcasing that modality-aware distillation complements experience replay. Overall, our results emphasize the importance of addressing modality-specific dynamics to prevent forgetting in multimodal continual learning.

3. Shaking Up VLMs: Comparing Transformers and Structured State Space Models for Vision&Language Modeling

   Georgios Pantazopoulos, Malvina Nikandrou, Alessandro Suglia, and 2 more authors

   In , Aug 2024

2023

1. Quality-agnostic image captioning to safely assist people with vision impairment

   Lu Yu, Malvina Nikandrou, Jiali Jin, and 1 more author

   In Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence, Aug 2023
2. EMMA: A Foundation Model for Embodied, Interactive, Multimodal Task Completion in 3D Environments

   Amit Parekh, Malvina Nikandrou, Georgios Pantazopoulos, and 5 more authors

   Alexa Prize SimBot Challenge Proceedings, Aug 2023
3. Multitask Multimodal Prompted Training for Interactive Embodied Task Completion

   Georgios Pantazopoulos, Malvina Nikandrou, Amit Parekh, and 6 more authors

   In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, Dec 2023

   Abs

   Interactive and embodied tasks pose at least two fundamental challenges to existing Vision & Language (VL) models, including 1) grounding language in trajectories of actions and observations, and 2) referential disambiguation. To tackle these challenges, we propose an Embodied MultiModal Agent (EMMA): a unified encoder-decoder model that reasons over images and trajectories, and casts action prediction as multimodal text generation. By unifying all tasks as text generation, EMMA learns a language of actions which facilitates transfer across tasks. Different to previous modular approaches with independently trained components, we use a single multitask model where each task contributes to goal completion. EMMA performs on par with similar models on several VL benchmarks and sets a new state-of-the-art performance (36.81% success rate) on the Dialog-guided Task Completion (DTC), a benchmark to evaluate dialog-guided agents in the Alexa Arena.

2022

1. Going for GOAL: A resource for grounded football commentaries

   Alessandro Suglia, José Lopes, Emanuele Bastianelli, and 6 more authors

   arXiv preprint arXiv:2211.04534, Dec 2022
2. Task Formulation Matters When Learning Continually: A Case Study in Visual Question Answering

   Mavina Nikandrou, Lu Yu, Alessandro Suglia, and 2 more authors

   Dec 2022

2021

1. ViCA: Combining visual, social, and task-oriented conversational AI in a healthcare setting

   George Pantazopoulos, Jeremy Bruyere, Malvina Nikandrou, and 6 more authors

   In Proceedings of the 2021 International Conference on Multimodal Interaction, Dec 2021