An official website of the United States government
Recent advances in Large Language Models (LLMs) have demonstrated significant potential in the field of Recommendation Systems (RSs). Most existing studies have focused on converting user behavior logs into textual prompts and leveraging techniques such as prompt tuning to enable LLMs for recommendation tasks. Meanwhile, research interest has recently grown in multimodal recommendation systems that integrate data from images, text, and other sources using modality fusion techniques. This introduces new challenges to the existing LLM-based recommendation paradigm which relies solely on text modality information. Moreover, although Multimodal Large Language Models (MLLMs) capable of processing multi-modal inputs have emerged, how to equip MLLMs with multi-modal recommendation capabilities remains largely unexplored. To this end, in this paper, we propose the Multimodal Large Language Model-enhanced Sequential Multimodal Recommendation (MLLM-MSR) model. To capture the dynamic user preference, we design a two-stage user preference summarization method. Specifically, we first utilize an MLLM-based item-summarizer to extract image feature given an item and convert the image into text. Then, we employ a recurrent user preference summarization generation paradigm to capture the dynamic changes in user preferences based on an LLM-based user-summarizer. Finally, to enable the MLLM for multi-modal recommendation task, we propose to fine-tune a MLLM-based recommender using Supervised Fine-Tuning (SFT) techniques. Extensive evaluations across various datasets validate the effectiveness of MLLM-MSR, showcasing its superior ability to capture and adapt to the evolving dynamics of user preferences.
more »
« less
Harmonizer: A Universal Signal Tokenization Framework for Multimodal Large Language Models
This paper introduces Harmonizer, a universal framework designed for tokenizing heterogeneous input signals, including text, audio, and video, to enable seamless integration into multimodal large language models (LLMs). Harmonizer employs a unified approach to convert diverse, non-linguistic signals into discrete tokens via its FusionQuantizer architecture, built on FluxFormer, to efficiently capture essential signal features while minimizing complexity. We enhance features through STFT-based spectral decomposition, Hilbert transform analytic signal extraction, and SCLAHE spectrogram contrast optimization, and train using a composite loss function to produce reliable embeddings and construct a robust vector vocabulary. Experimental validation on music datasets such as E-GMD v1.0.0, Maestro v3.0.0, and GTZAN demonstrates high fidelity across 288 s of vocal signals (MSE = 0.0037, CC = 0.9282, Cosine Sim. = 0.9278, DTW = 12.12, MFCC Sim. = 0.9997, Spectral Conv. = 0.2485). Preliminary tests on text reconstruction and UCF-101 video clips further confirm Harmonizer’s applicability across discrete and spatiotemporal modalities. Rooted in the universality of wave phenomena and Fourier theory, Harmonizer offers a physics-inspired, modality-agnostic fusion mechanism via wave superposition and interference principles. In summary, Harmonizer integrates natural language processing and signal processing into a coherent tokenization paradigm for efficient, interpretable multimodal learning.
more »
« less
- Award ID(s):
- 1924278
- PAR ID:
- 10661020
- Publisher / Repository:
- MDPI
- Date Published:
- Journal Name:
- Mathematics
- Volume:
- 13
- Issue:
- 11
- ISSN:
- 2227-7390
- Page Range / eLocation ID:
- 1819
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Millimeter-wave (mmWave) sensing has emerged as a promising technology for non-contact health monitoring, offering high spatial resolution, material sensitivity, and integration potential with wireless platforms. While prior work has focused on specific applications or signal processing methods, a unified understanding of how mmWave signals map to clinically relevant biomarkers remains lacking. This survey presents a full-stack review of mmWave-based medical sensing systems, encompassing signal acquisition, physical feature extraction, modeling strategies, and potential medical and healthcare uses. We introduce a taxonomy that decouples low-level mmWave signal features—such as motion, material property, and structure—from high-level biomedical biomarkers, including respiration pattern, heart rate, tissue hydration, and gait. We then classify and contrast the modeling approaches—ranging from physics-driven analytical models to machine learning techniques—that enable this mapping. Furthermore, we analyze representative studies across vital signs monitoring, cardiovascular assessment, wound evaluation, and neuro-motor disorders. By bridging wireless sensing and medical interpretation, this work offers a structured reference for designing next-generation mmWave health monitoring systems. We conclude by discussing open challenges, including model interpretability, clinical validation, and multimodal integration.more » « less
-
Group meetings can suffer from serious problems that undermine performance, including bias, "groupthink", fear of speaking, and unfocused discussion. To better understand these issues, propose interventions, and thus improve team performance, we need to study human dynamics in group meetings. However, this process currently heavily depends on manual coding and video cameras. Manual coding is tedious, inaccurate, and subjective, while active video cameras can affect the natural behavior of meeting participants. Here, we present a smart meeting room that combines microphones and unobtrusive ceiling-mounted Time-of-Flight (ToF) sensors to understand group dynamics in team meetings. We automatically process the multimodal sensor outputs with signal, image, and natural language processing algorithms to estimate participant head pose, visual focus of attention (VFOA), non-verbal speech patterns, and discussion content. We derive metrics from these automatic estimates and correlate them with user-reported rankings of emergent group leaders and major contributors to produce accurate predictors. We validate our algorithms and report results on a new dataset of lunar survival tasks of 36 individuals across 10 groups collected in the multimodal-sensor-enabled smart room.more » « less
-
Generative models such as Large Language Models (LLM) and Multimodal Large Language models (MLLMs) trained on massive web corpora can memorize and disclose individuals’ confidential and private data, raising legal and ethical concerns. While many previous works have addressed this issue in LLM via machine unlearning, it remains largely unexplored for MLLMs. To tackle this challenge, we introduce Multimodal Large Language Model Unlearning Benchmark (MLLMU-Bench), a novel benchmark aimed at advancing the understanding of multimodal machine unlearning. MLLMU-Bench consists of 500 fictitious profiles and 153 profiles for public celebrities, each profile feature over 14 customized question-answer pairs, evaluated from both multimodal (image+text) and unimodal (text) perspectives. The benchmark is divided into four sets to assess unlearning algorithms in terms of efficacy, generalizability, and model utility. Finally, we provide baseline results using existing generative model unlearning algorithms. Surprisingly, our experiments show that unimodal unlearning algorithms excel in generation tasks, while multimodal unlearning approaches perform better in classification with multimodal inputs.more » « less
-
Grounded language acquisition is a major area of research combining aspects of natural language processing, computer vision, and signal processing, compounded by domain issues requiring sample efficiency and other deployment constraints. In this work, we present a multimodal dataset of RGB+depth objects with spoken as well as textual descriptions. We analyze the differences between the two types of descriptive language and our experiments demonstrate that the different modalities affect learning. This will enable researchers studying the intersection of robotics, NLP, and HCI to better investigate how the multiple modalities of image, depth, text, speech, and transcription interact, as well as how differences in the vernacular of these modalities impact results.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript
- Journal Article:
- https://doi.org/10.3390/math13111819
