Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 20 of 37 for “"MLLMs"”.
-
Grounding Semantics and Instructions with Vision
… the latest Multimodal Large Language Models (MLLMs) shows superb visual instruction following capability for multimodal tasks, however, the existing MLLMs heavily reply on the pretrained LLMs, inherit its language priors for instruction understanding. The third part of this thesis aims to …
-
How Well Do Multimodal Large Language Models Score on Visual Personnel Assessments
Multimodal Large Language Models (MLLMs) pose new threats to the validity of visual personnel assessments in high stakes selection contexts, as their emergent visual perception and understanding capabilities may facilitate applicant cheating. The study investigated the performance of three popular …
-
From Multimodal LLMs to Generalist Embodied Agents
… how finetuning multimodal large language models (MLLMs) with large-scale embodied experience unlocks their ability to act as generalist embodied agents. While MLLMs provide reasoning abilities and broad world knowledge, finetuning aligns these capabilities with the demands of embodied agents, …
-
Belonging: Stories from Multilingual Learners in the Linguistic Minority
… learners in the linguistic minority (MLLMs) experience connection and belonging in a Rocky Mountain middle school. Although multilingual learners (MLs) have received increased attention, MLLMs, whose home languages are not widely spoken in the school context, remain overlooked. …
-
Hallucinations in Multimodal Large Language Models: Evaluation, Mitigation, and Future Directions
Multimodal Large Language Models (MLLMs) have achieved impressive performance across a wide array of tasks. Despite these promising developments, MLLMs often generate outputs that are inconsistent with the visual content, a challenge known as hallucination, which poses substantial obstacles to …
-
Video Understanding with Attention Encoder and Multimodal Large Language Model
… emergence of Multimodal Large Language Models (MLLMs). While MLLMs have demonstrated significant promise, effectively capturing and reasoning about complex temporal dynamics and object-level interactions in videos remains an active area of research. This project introduces a novel framework …
-
Towards trustworthy multimodal large language models
… advances in multimodal large language models (MLLMs) have yielded remarkable capabilities in both visual understanding and content generation~\citep{liu2023llava,gpt4v,wu2024janus,openai2024gpt4ocard}. However, these progresses are accompanied by critical challenges in the domains of …
-
Extending Multimodal Large Language Models Beyond Single Modalities
Multimodal large language models (MLLMs) aim to unify diverse sensory inputs—such as text, vision, and audio—within a single reasoning framework, enabling more comprehensive and human-like understanding. However, as the number of supported modalities increases, generalist MLLMs often underperform …
-
DATA-EFFICIENT TRAINING OF MULTIMODAL LARGE LANGUAGE MODELS THROUGH FINE-GRAINED SELECTION AND MODALITY ANALYSIS
… for aligning multimodal large language models (MLLMs) with human intent, but its effectiveness is often hindered by noisy and redundant training data. In this work, we propose ModFilter, a fine-grained, forward-only data selection method designed for MLLMs. Unlike prior approaches that operate …
-
Guidance Conditions in Generative Modeling: Elevating Discriminative Capabilities and Controllability
… accuracy in multi-modal large language models (MLLMs) through refined multimodal conditioning. This work introduces a self-visual-RAG framework that enables MLLMs to process multi-page, visually rich documents efficiently. It optimizes MLLMs as multimodal retrievers, reducing the input length by …
-
Submodular Data Selection and Augmentation for Resource-Efficient Learning
… (ViTs), and Multimodal Large Language Models (MLLMs), require vast datasets, substantial memory, and extensive compute resources, making them difficult to scale and deploy—especially in resource-constrained and data-scarce environments. These challenges are further compounded by the redundancy …
-
PREDICTING COLLECTIVE VIOLENCE FROM COORDINATED HOSTILE INFORMATION CAMPAIGNS IN SOCIAL MEDIA
… learning classifiers against deep learning MLLMs for predicting message classes linked to past and future occurrences of violent events. Our findings suggest that MLLMs substantially outperform traditional ML models in predictive accuracy. One major contribution of our work is that military …
-
Generalizable Adaptation for Vision-Language Models
… (VLMs) and Multimodal Large Language Models (MLLMs) have recently emerged as powerful frameworks for learning joint representations across visual and textual modalities. These models enable a wide range of applications, including visual recognition, multimodal reasoning, and visual question …
-
LLM-Powered Question Answering for Object States in Virtual Reality
… (LLMs) and multimodal large language models (MLLMs) enable natural language–based querying in virtual reality (VR). However, VR environments are highly localized, personalized, and dynamic, making it challenging for general-purpose models to answer environment-specific queries or reason about …
-
Towards an AI-Enabled Metaverse: Architecture, Resource Allocation, and Multimodal Benchmarking
… of Multimodal Large Language Models (MLLMs). Thirteen representative MLLMs are assessed through a pairwise comparison protocol under an MLLM-as-a-judge paradigm guided by metaverse-specific semantic criteria. Human expert validation confirms the robustness of the proposed benchmark, …
-
Transferable and Scalable Vision-Language Models for Reasoning and Grounding
… training of multimodal large language models (MLLMs) for fine-grained image and video understanding. We begin with VistaLLM, a unified framework that addresses coarse- and fine-grained vision language tasks across single and multiple photos. VistaLLM is followed by ED-VTG, a two-stage approach …
-
Data-Efficient Learning for Robust Vision and Multimodal Models in Open-World Scenarios
… (VIT) of multimodal large language models (MLLMs), a framework that filters images prior to instruction generation. This approach significantly reduces redundancy in instruction tuning and generation (e.g., API usage) for MLLMs, enabling efficient fine-tuning without sacrificing performance. …
-
Improving Vulnerable Road Users' Safety Through Computer Vision-Based Crossing Direction Prediction And Multimodal Large Language Model-Based Accident Scene Description
… assessment of multimodal large language models (MLLMs) in understanding accident scenarios involving VRUs. First, the proposed framework leverages multi-modal cues, including 3D pose estimation and spatio-temporal trajectories, to predict VRU crossing intentions at intersections. By combining a …
-
Researching an enhanced multimodal learning framework for improved inter-modal and intra-modal alignment
… As the size of Multimodal Large Language Models (MLLMs) increases, their adaptation to specific tasks becomes more complex, primarily due to heightened computational and memory requirements. Traditional fine-tuning methods, while effective, are resource-intensive and necessitate extensive, …
-
TOWARD HUMAN-CENTRIC VISUAL INTELLIGENCE THROUGH MULTIMODAL LEARNING
… expectations. Multimodal Large Language Models (MLLMs) provide rich semantic priors, yet remain limited in spatial grounding, computational efficiency, and reliable preference alignment. This dissertation addresses these challenges through multimodal methods that connect structured perception, …
Page 1 of 2