Instruction tuned Large Vision Language Models (LVLMs) have made significant advancements in generalizing across a diverse set of multimodal tasks, especially for Visual Question Answering (VQA). However, generating detailed responses that are visually grounded is still a challenging task for these models. We find that even the current state-of-the-art LVLMs (InstructBLIP) still contain a staggering 30 percent of hallucinatory text in the form of non-existent objects, unfaithful descriptions, and inaccurate relationships. To address this, we introduce M-HalDetect, a {M}ultimodal {Hal}lucination {Detect}ion Dataset that can be used to train and benchmark models for hallucination detection and prevention. M-HalDetect consists of 16k fine-grained labels on VQA examples, making it the first comprehensive multi-modal hallucination detection dataset for detailed image descriptions. Unlike previous work that only consider object hallucination, we additionally annotate both entity descriptions and relationships that are unfaithful. To demonstrate the potential of this dataset for preference alignment, we propose fine-grained Direct Preference Optimization, as well as train fine-grained multi-modal reward models and evaluate their effectiveness with best-of-n rejection sampling. We perform human evaluation on both DPO and rejection sampling, and find that they reduce hallucination rates by 41% and 55% respectively, a significant improvement over the baseline.
Consider the spiked Wigner model \[ X = \sum_{i = 1}^k \lambda_i u_i u_i^\top + \sigma G, \] where $G$ is an $N \times N$ GOE random matrix, and the eigenvalues $\lambda_i$ are all spiked, i.e. above the Baik-Ben Arous-P\'ech\'e (BBP) threshold $\sigma$. We consider AIC-type model selection criteria of the form \[ -2 \, (\text{maximised log-likelihood}) + \gamma \, (\text{number of parameters}) \] for estimating the number $k$ of spikes. For $\gamma > 2$, the above criterion is strongly consistent provided $\lambda_k > \lambda_{\gamma}$, where $\lambda_{\gamma}$ is a threshold strictly above the BBP threshold, whereas for $\gamma < 2$, it almost surely overestimates $k$. Although AIC (which corresponds to $\gamma = 2$) is not strongly consistent, we show that taking $\gamma = 2 + \delta_N$, where $\delta_N \to 0$ and $\delta_N \gg N^{-2/3}$, results in a weakly consistent estimator of $k$. We also show that a certain soft minimiser of AIC is strongly consistent.
Prosthetic Joint Infection (PJI) is a prevalent and severe complication characterized by high diagnostic challenges. Currently, a unified diagnostic standard incorporating both computed tomography (CT) images and numerical text data for PJI remains unestablished, owing to the substantial noise in CT images and the disparity in data volume between CT images and text data. This study introduces a diagnostic method, HGT, based on deep learning and multimodal techniques. It effectively merges features from CT scan images and patients' numerical text data via a Unidirectional Selective Attention (USA) mechanism and a graph convolutional network (GCN)-based feature fusion network. We evaluated the proposed method on a custom-built multimodal PJI dataset, assessing its performance through ablation experiments and interpretability evaluations. Our method achieved an accuracy (ACC) of 91.4\% and an area under the curve (AUC) of 95.9\%, outperforming recent multimodal approaches by 2.9\% in ACC and 2.2\% in AUC, with a parameter count of only 68M. Notably, the interpretability results highlighted our model's strong focus and localization capabilities at lesion sites. This proposed method could provide clinicians with additional diagnostic tools to enhance accuracy and efficiency in clinical practice.
Cross-lingual open information extraction aims to extract structured information from raw text across multiple languages. Previous work uses a shared cross-lingual pre-trained model to handle the different languages but underuses the potential of the language-specific representation. In this paper, we propose an effective multi-stage tuning framework called MT4CrossIE, designed for enhancing cross-lingual open information extraction by injecting language-specific knowledge into the shared model. Specifically, the cross-lingual pre-trained model is first tuned in a shared semantic space (e.g., embedding matrix) in the fixed encoder and then other components are optimized in the second stage. After enough training, we freeze the pre-trained model and tune the multiple extra low-rank language-specific modules using mixture-of-LoRAs for model-based cross-lingual transfer. In addition, we leverage two-stage prompting to encourage the large language model (LLM) to annotate the multi-lingual raw data for data-based cross-lingual transfer. The model is trained with multi-lingual objectives on our proposed dataset OpenIE4++ by combing the model-based and data-based transfer techniques. Experimental results on various benchmarks emphasize the importance of aggregating multiple plug-in-and-play language-specific modules and demonstrate the effectiveness of MT4CrossIE in cross-lingual OIE\footnote{\url{https://github.com/CSJianYang/Multilingual-Multimodal-NLP}}.
Multi-modal recommendation systems, which integrate diverse types of information, have gained widespread attention in recent years. However, compared to traditional collaborative filtering-based multi-modal recommendation systems, research on multi-modal sequential recommendation is still in its nascent stages. Unlike traditional sequential recommendation models that solely rely on item identifier (ID) information and focus on network structure design, multi-modal recommendation models need to emphasize item representation learning and the fusion of heterogeneous data sources. This paper investigates the impact of item representation learning on downstream recommendation tasks and examines the disparities in information fusion at different stages. Empirical experiments are conducted to demonstrate the need to design a framework suitable for collaborative learning and fusion of diverse information. Based on this, we propose a new model-agnostic framework for multi-modal sequential recommendation tasks, called Online Distillation-enhanced Multi-modal Transformer (ODMT), to enhance feature interaction and mutual learning among multi-source input (ID, text, and image), while avoiding conflicts among different features during training, thereby improving recommendation accuracy. To be specific, we first introduce an ID-aware Multi-modal Transformer module in the item representation learning stage to facilitate information interaction among different features. Secondly, we employ an online distillation training strategy in the prediction optimization stage to make multi-source data learn from each other and improve prediction robustness. Experimental results on a stream media recommendation dataset and three e-commerce recommendation datasets demonstrate the effectiveness of the proposed two modules, which is approximately 10% improvement in performance compared to baseline models.
Zero-shot video recognition (ZSVR) is a task that aims to recognize video categories that have not been seen during the model training process. Recently, vision-language models (VLMs) pre-trained on large-scale image-text pairs have demonstrated impressive transferability for ZSVR. To make VLMs applicable to the video domain, existing methods often use an additional temporal learning module after the image-level encoder to learn the temporal relationships among video frames. Unfortunately, for video from unseen categories, we observe an abnormal phenomenon where the model that uses spatial-temporal feature performs much worse than the model that removes temporal learning module and uses only spatial feature. We conjecture that improper temporal modeling on video disrupts the spatial feature of the video. To verify our hypothesis, we propose Feature Factorization to retain the orthogonal temporal feature of the video and use interpolation to construct refined spatial-temporal feature. The model using appropriately refined spatial-temporal feature performs better than the one using only spatial feature, which verifies the effectiveness of the orthogonal temporal feature for the ZSVR task. Therefore, an Orthogonal Temporal Interpolation module is designed to learn a better refined spatial-temporal video feature during training. Additionally, a Matching Loss is introduced to improve the quality of the orthogonal temporal feature. We propose a model called OTI for ZSVR by employing orthogonal temporal interpolation and the matching loss based on VLMs. The ZSVR accuracies on popular video datasets (i.e., Kinetics-600, UCF101 and HMDB51) show that OTI outperforms the previous state-of-the-art method by a clear margin.
Automatic medication mining from clinical and biomedical text has become a popular topic due to its real impact on healthcare applications and the recent development of powerful language models (LMs). However, fully-automatic extraction models still face obstacles to be overcome such that they can be deployed directly into clinical practice for better impacts. Such obstacles include their imbalanced performances on different entity types and clinical events. In this work, we examine current state-of-the-art pre-trained language models (PLMs) on such tasks, via fine-tuning including the monolingual model Med7 and multilingual large language model (LLM) XLM-RoBERTa. We compare their advantages and drawbacks using historical medication mining shared task data sets from n2c2-2018 challenges. We report the findings we get from these fine-tuning experiments such that they can facilitate future research on addressing them, for instance, how to combine their outputs, merge such models, or improve their overall accuracy by ensemble learning and data augmentation. MedMine is part of the M3 Initiative \url{https://github.com/HECTA-UoM/M3}
We introduce text2fabric, a novel dataset that links free-text descriptions to various fabric materials. The dataset comprises 15,000 natural language descriptions associated to 3,000 corresponding images of fabric materials. Traditionally, material descriptions come in the form of tags/keywords, which limits their expressivity, induces pre-existing knowledge of the appropriate vocabulary, and ultimately leads to a chopped description system. Therefore, we study the use of free-text as a more appropriate way to describe material appearance, taking the use case of fabrics as a common item that non-experts may often deal with. Based on the analysis of the dataset, we identify a compact lexicon, set of attributes and key structure that emerge from the descriptions. This allows us to accurately understand how people describe fabrics and draw directions for generalization to other types of materials. We also show that our dataset enables specializing large vision-language models such as CLIP, creating a meaningful latent space for fabric appearance, and significantly improving applications such as fine-grained material retrieval and automatic captioning.
The General Transit Feed Specification (GTFS) standard for publishing transit data is ubiquitous. GTFS being tabular data, with information spread across different files, necessitates specialized tools or packages to retrieve information. Concurrently, the use of Large Language Models for text and information retrieval is growing. The idea of this research is to see if the current widely adopted LLMs (ChatGPT) are able to retrieve information from GTFS using natural language instructions. We first test whether ChatGPT (GPT-3.5) understands the GTFS specification. GPT-3.5 answers 77% of our multiple-choice questions (MCQ) correctly. Next, we task the LLM with information extractions from a filtered GTFS feed with 4 routes. For information retrieval, we compare zero-shot and program synthesis. Program synthesis works better, achieving ~90% accuracy on simple questions and ~40% accuracy on complex questions.
Text-to-speech(TTS) has undergone remarkable improvements in performance, particularly with the advent of Denoising Diffusion Probabilistic Models (DDPMs). However, the perceived quality of audio depends not solely on its content, pitch, rhythm, and energy, but also on the physical environment. In this work, we propose ViT-TTS, the first visual TTS model with scalable diffusion transformers. ViT-TTS complement the phoneme sequence with the visual information to generate high-perceived audio, opening up new avenues for practical applications of AR and VR to allow a more immersive and realistic audio experience. To mitigate the data scarcity in learning visual acoustic information, we 1) introduce a self-supervised learning framework to enhance both the visual-text encoder and denoiser decoder; 2) leverage the diffusion transformer scalable in terms of parameters and capacity to learn visual scene information. Experimental results demonstrate that ViT-TTS achieves new state-of-the-art results, outperforming cascaded systems and other baselines regardless of the visibility of the scene. With low-resource data (1h, 2h, 5h), ViT-TTS achieves comparative results with rich-resource baselines.~\footnote{Audio samples are available at \url{https://ViT-TTS.github.io/.}}