Multimodal deep learning, especially vision-language models, have gained significant traction in recent years, greatly improving performance on many downstream tasks, including content moderation and violence detection. However, standard multimodal approaches often assume consistent modalities between training and inference, limiting applications in many real-world use cases, as some modalities may not be available during inference. While existing research mitigates this problem through reconstructing the missing modalities, they unavoidably increase unnecessary computational cost, which could be just as critical, especially for large, deployed infrastructures in industry. To this end, we propose a novel guidance network that promotes knowledge sharing during training, taking advantage of the multimodal representations to train better single-modality models to be used for inference. Real-world experiments in violence detection shows that our proposed framework trains single-modality models that significantly outperform traditionally trained counterparts, while avoiding increases in computational cost for inference.
Multimodal deep learning foundation models can learn the relationship between images and text. In the context of medical imaging, mapping images to language concepts reflects the clinical task of diagnostic image interpretation, however current general-purpose foundation models do not perform well in this context because their training corpus have limited medical text and images. To address this challenge and account for the range of cardiac physiology, we leverage 1,032,975 cardiac ultrasound videos and corresponding expert interpretations to develop EchoCLIP, a multimodal foundation model for echocardiography. EchoCLIP displays strong zero-shot (not explicitly trained) performance in cardiac function assessment (external validation left ventricular ejection fraction mean absolute error (MAE) of 7.1%) and identification of implanted intracardiac devices (areas under the curve (AUC) between 0.84 and 0.98 for pacemakers and artificial heart valves). We also developed a long-context variant (EchoCLIP-R) with a custom echocardiography report text tokenizer which can accurately identify unique patients across multiple videos (AUC of 0.86), identify clinical changes such as orthotopic heart transplants (AUC of 0.79) or cardiac surgery (AUC 0.77), and enable robust image-to-text search (mean cross-modal retrieval rank in the top 1% of candidate text reports). These emergent capabilities can be used for preliminary assessment and summarization of echocardiographic findings.
Multi-modal foundation models combining vision and language models such as Flamingo or GPT-4 have recently gained enormous interest. Alignment of foundation models is used to prevent models from providing toxic or harmful output. While malicious users have successfully tried to jailbreak foundation models, an equally important question is if honest users could be harmed by malicious third-party content. In this paper we show that imperceivable attacks on images in order to change the caption output of a multi-modal foundation model can be used by malicious content providers to harm honest users e.g. by guiding them to malicious websites or broadcast fake information. This indicates that countermeasures to adversarial attacks should be used by any deployed multi-modal foundation model.
In a joint vision-language space, a text feature (e.g., from "a photo of a dog") could effectively represent its relevant image features (e.g., from dog photos). Also, a recent study has demonstrated the cross-modal transferability phenomenon of this joint space. From these observations, we propose PromptStyler which simulates various distribution shifts in the joint space by synthesizing diverse styles via prompts without using any images to deal with source-free domain generalization. The proposed method learns to generate a variety of style features (from "a S* style of a") via learnable style word vectors for pseudo-words S*. To ensure that learned styles do not distort content information, we force style-content features (from "a S* style of a [class]") to be located nearby their corresponding content features (from "[class]") in the joint vision-language space. After learning style word vectors, we train a linear classifier using synthesized style-content features. PromptStyler achieves the state of the art on PACS, VLCS, OfficeHome and DomainNet, even though it does not require any images for training.
Text-embedded images can serve as a means of spreading hate speech, propaganda, and extremist beliefs. Throughout the Russia-Ukraine war, both opposing factions heavily relied on text-embedded images as a vehicle for spreading propaganda and hate speech. Ensuring the effective detection of hate speech and propaganda is of utmost importance to mitigate the negative effect of hate speech dissemination. In this paper, we outline our methodologies for two subtasks of Multimodal Hate Speech Event Detection 2023. For the first subtask, hate speech detection, we utilize multimodal deep learning models boosted by ensemble learning and syntactical text attributes. For the second subtask, target detection, we employ multimodal deep learning models boosted by named entity features. Through experimentation, we demonstrate the superior performance of our models compared to all textual, visual, and text-visual baselines employed in multimodal hate speech detection. Furthermore, our models achieve the first place in both subtasks on the final leaderboard of the shared task.
Computer-assisted diagnostic and prognostic systems of the future should be capable of simultaneously processing multimodal data. Multimodal deep learning (MDL), which involves the integration of multiple sources of data, such as images and text, has the potential to revolutionize the analysis and interpretation of biomedical data. However, it only caught researchers' attention recently. To this end, there is a critical need to conduct a systematic review on this topic, identify the limitations of current work, and explore future directions. In this scoping review, we aim to provide a comprehensive overview of the current state of the field and identify key concepts, types of studies, and research gaps with a focus on biomedical images and texts joint learning, mainly because these two were the most commonly available data types in MDL research. This study reviewed the current uses of multimodal deep learning on five tasks: (1) Report generation, (2) Visual question answering, (3) Cross-modal retrieval, (4) Computer-aided diagnosis, and (5) Semantic segmentation. Our results highlight the diverse applications and potential of MDL and suggest directions for future research in the field. We hope our review will facilitate the collaboration of natural language processing (NLP) and medical imaging communities and support the next generation of decision-making and computer-assisted diagnostic system development.
Over the last decade, there has been a vast increase in eating disorder diagnoses and eating disorder-attributed deaths, reaching their zenith during the Covid-19 pandemic. This immense growth derived in part from the stressors of the pandemic but also from increased exposure to social media, which is rife with content that promotes eating disorders. This study aimed to create a multimodal deep learning model that can determine if a given social media post promotes eating disorders based on a combination of visual and textual data. A labeled dataset of Tweets was collected from Twitter, recently rebranded as X, upon which twelve deep learning models were trained and evaluated. Based on model performance, the most effective deep learning model was the multimodal fusion of the RoBERTa natural language processing model and the MaxViT image classification model, attaining accuracy and F1 scores of 95.9% and 0.959, respectively. The RoBERTa and MaxViT fusion model, deployed to classify an unlabeled dataset of posts from the social media sites Tumblr and Reddit, generated results akin to those of previous research studies that did not employ artificial intelligence-based techniques, indicating that deep learning models can develop insights congruent to those of researchers. Additionally, the model was used to conduct a time-series analysis of yet unseen Tweets from eight Twitter hashtags, uncovering that, since 2014, the relative abundance of content that promotes eating disorders has decreased drastically within those communities. Despite this reduction, by 2018, content that promotes eating disorders had either stopped declining or increased in ampleness anew on those hashtags.
Renal cell carcinoma represents a significant global health challenge with a low survival rate. This research aimed to devise a comprehensive deep-learning model capable of predicting survival probabilities in patients with renal cell carcinoma by integrating CT imaging and clinical data and addressing the limitations observed in prior studies. The aim is to facilitate the identification of patients requiring urgent treatment. The proposed framework comprises three modules: a 3D image feature extractor, clinical variable selection, and survival prediction. The feature extractor module, based on the 3D CNN architecture, predicts the ISUP grade of renal cell carcinoma tumors linked to mortality rates from CT images. A selection of clinical variables is systematically chosen using the Spearman score and random forest importance score as criteria. A deep learning-based network, trained with discrete LogisticHazard-based loss, performs the survival prediction. Nine distinct experiments are performed, with varying numbers of clinical variables determined by different thresholds of the Spearman and importance scores. Our findings demonstrate that the proposed strategy surpasses the current literature on renal cancer prognosis based on CT scans and clinical factors. The best-performing experiment yielded a concordance index of 0.84 and an area under the curve value of 0.8 on the test cohort, which suggests strong predictive power. The multimodal deep-learning approach developed in this study shows promising results in estimating survival probabilities for renal cell carcinoma patients using CT imaging and clinical data. This may have potential implications in identifying patients who require urgent treatment, potentially improving patient outcomes. The code created for this project is available for the public on: \href{https://github.com/Balasingham-AI-Group/Survival_CTplusClinical}{GitHub}
Learning multimodal representations involves integrating information from multiple heterogeneous sources of data. In order to accelerate progress towards understudied modalities and tasks while ensuring real-world robustness, we release MultiZoo, a public toolkit consisting of standardized implementations of > 20 core multimodal algorithms and MultiBench, a large-scale benchmark spanning 15 datasets, 10 modalities, 20 prediction tasks, and 6 research areas. Together, these provide an automated end-to-end machine learning pipeline that simplifies and standardizes data loading, experimental setup, and model evaluation. To enable holistic evaluation, we offer a comprehensive methodology to assess (1) generalization, (2) time and space complexity, and (3) modality robustness. MultiBench paves the way towards a better understanding of the capabilities and limitations of multimodal models, while ensuring ease of use, accessibility, and reproducibility. Our toolkits are publicly available, will be regularly updated, and welcome inputs from the community.
The robustness of multimodal deep learning models to realistic changes in the input text is critical for their applicability to important tasks such as text-to-image retrieval and cross-modal entailment. To measure robustness, several existing approaches edit the text data, but do so without leveraging the cross-modal information present in multimodal data. Information from the visual modality, such as color, size, and shape, provide additional attributes that users can include in their inputs. Thus, we propose cross-modal attribute insertions as a realistic perturbation strategy for vision-and-language data that inserts visual attributes of the objects in the image into the corresponding text (e.g., "girl on a chair" to "little girl on a wooden chair"). Our proposed approach for cross-modal attribute insertions is modular, controllable, and task-agnostic. We find that augmenting input text using cross-modal insertions causes state-of-the-art approaches for text-to-image retrieval and cross-modal entailment to perform poorly, resulting in relative drops of 15% in MRR and 20% in $F_1$ score, respectively. Crowd-sourced annotations demonstrate that cross-modal insertions lead to higher quality augmentations for multimodal data than augmentations using text-only data, and are equivalent in quality to original examples. We release the code to encourage robustness evaluations of deep vision-and-language models: https://github.com/claws-lab/multimodal-robustness-xmai.
Training multimodal networks requires a vast amount of data due to their larger parameter space compared to unimodal networks. Active learning is a widely used technique for reducing data annotation costs by selecting only those samples that could contribute to improving model performance. However, current active learning strategies are mostly designed for unimodal tasks, and when applied to multimodal data, they often result in biased sample selection from the dominant modality. This unfairness hinders balanced multimodal learning, which is crucial for achieving optimal performance. To address this issue, we propose three guidelines for designing a more balanced multimodal active learning strategy. Following these guidelines, a novel approach is proposed to achieve more fair data selection by modulating the gradient embedding with the dominance degree among modalities. Our studies demonstrate that the proposed method achieves more balanced multimodal learning by avoiding greedy sample selection from the dominant modality. Our approach outperforms existing active learning strategies on a variety of multimodal classification tasks. Overall, our work highlights the importance of balancing sample selection in multimodal active learning and provides a practical solution for achieving more balanced active learning for multimodal classification.
We present ImageBind, an approach to learn a joint embedding across six different modalities - images, text, audio, depth, thermal, and IMU data. We show that all combinations of paired data are not necessary to train such a joint embedding, and only image-paired data is sufficient to bind the modalities together. ImageBind can leverage recent large scale vision-language models, and extends their zero-shot capabilities to new modalities just by using their natural pairing with images. It enables novel emergent applications 'out-of-the-box' including cross-modal retrieval, composing modalities with arithmetic, cross-modal detection and generation. The emergent capabilities improve with the strength of the image encoder and we set a new state-of-the-art on emergent zero-shot recognition tasks across modalities, outperforming specialist supervised models. Finally, we show strong few-shot recognition results outperforming prior work, and that ImageBind serves as a new way to evaluate vision models for visual and non-visual tasks.
This work aims to create a multimodal AI system that chats with humans and shares relevant photos. While earlier works were limited to dialogues about specific objects or scenes within images, recent works have incorporated images into open-domain dialogues. However, their response generators are unimodal, accepting text input but no image input, thus prone to generating responses contradictory to the images shared in the dialogue. Therefore, this work proposes a complete chatbot system using two multimodal deep learning models: an image retriever that understands texts and a response generator that understands images. The image retriever, implemented by ViT and BERT, selects the most relevant image given the dialogue history and a database of images. The response generator, implemented by ViT and GPT-2/DialoGPT, generates an appropriate response given the dialogue history and the most recently retrieved image. The two models are trained and evaluated on PhotoChat, an open-domain dialogue dataset in which a photo is shared in each session. In automatic evaluation, the proposed image retriever outperforms existing baselines VSE++ and SCAN with Recall@1/5/10 of 0.1/0.3/0.4 and MRR of 0.2 when ranking 1,000 images. The proposed response generator also surpasses the baseline Divter with PPL of 16.9, BLEU-1/2 of 0.13/0.03, and Distinct-1/2 of 0.97/0.86, showing a significant improvement in PPL by -42.8 and BLEU-1/2 by +0.07/0.02. In human evaluation with a Likert scale of 1-5, the complete multimodal chatbot system receives higher image-groundedness of 4.3 and engagingness of 4.3, along with competitive fluency of 4.1, coherence of 3.9, and humanness of 3.1, when compared to other chatbot variants. The source code is available at: https://github.com/minniie/multimodal_chat.git.
In recent years, AI models that mine intrinsic patterns from molecular structures and protein sequences have shown promise in accelerating drug discovery. However, these methods partly lag behind real-world pharmaceutical approaches of human experts that additionally grasp structured knowledge from knowledge bases and unstructured knowledge from biomedical literature. To bridge this gap, we propose KEDD, a unified, end-to-end, and multimodal deep learning framework that optimally incorporates both structured and unstructured knowledge for vast AI drug discovery tasks. The framework first extracts underlying characteristics from heterogeneous inputs, and then applies multimodal fusion for accurate prediction. To mitigate the problem of missing modalities, we leverage multi-head sparse attention and a modality masking mechanism to extract relevant information robustly. Benefiting from integrated knowledge, our framework achieves a deeper understanding of molecule entities, brings significant improvements over state-of-the-art methods on a wide range of tasks and benchmarks, and reveals its promising potential in assisting real-world drug discovery.
The rise in loosely-structured data available through text, images, and other modalities has called for new ways of querying them. Multimedia Information Retrieval has filled this gap and has witnessed exciting progress in recent years. Tasks such as search and retrieval of extensive multimedia archives have undergone massive performance improvements, driven to a large extent by recent developments in multimodal deep learning. However, methods in this field remain limited in the kinds of queries they support and, in particular, their inability to answer database-like queries. For this reason, inspired by recent work on neural databases, we propose a new framework, which we name Multimodal Neural Databases (MMNDBs). MMNDBs can answer complex database-like queries that involve reasoning over different input modalities, such as text and images, at scale. In this paper, we present the first architecture able to fulfill this set of requirements and test it with several baselines, showing the limitations of currently available models. The results show the potential of these new techniques to process unstructured data coming from different modalities, paving the way for future research in the area. Code to replicate the experiments will be released at https://github.com/GiovanniTRA/MultimodalNeuralDatabases
Growing heterogeneity and configurability in HPC architectures has made auto-tuning applications and runtime parameters on these systems very complex. Users are presented with a multitude of options to configure parameters. In addition to application specific solutions, a common approach is to use general purpose search strategies, which often might not identify the best configurations or their time to convergence is a significant barrier. There is, thus, a need for a general purpose and efficient tuning approach that can be easily scaled and adapted to various tuning tasks. We propose a technique for tuning parallel code regions that is general enough to be adapted to multiple tasks. In this paper, we analyze IR-based programming models to make task-specific performance optimizations. To this end, we propose the Multimodal Graph Neural Network and Autoencoder (MGA) tuner, a multimodal deep learning based approach that adapts Heterogeneous Graph Neural Networks and Denoizing Autoencoders for modeling IR-based code representations that serve as separate modalities. This approach is used as part of our pipeline to model a syntax, semantics, and structure-aware IR-based code representation for tuning parallel code regions/kernels. We extensively experiment on OpenMP and OpenCL code regions/kernels obtained from PolyBench, Rodinia, STREAM, DataRaceBench, AMD SDK, NPB, NVIDIA SDK, Parboil, SHOC, and LULESH benchmarks. We apply our multimodal learning techniques to the tasks of i) optimizing the number of threads, scheduling policy and chunk size in OpenMP loops and, ii) identifying the best device for heterogeneous device mapping of OpenCL kernels. Our experiments show that this multimodal learning based approach outperforms the state-of-the-art in all experiments.
Medical tabular data, abundant in Electronic Health Records (EHRs), is a valuable resource for diverse medical tasks such as risk prediction. While deep learning approaches, particularly transformer-based models, have shown remarkable performance in tabular data prediction, there are still problems remained for existing work to be effectively adapted into medical domain, such as under-utilization of unstructured free-texts, limited exploration of textual information in structured data, and data corruption. To address these issues, we propose P-Transformer, a Prompt-based multimodal Transformer architecture designed specifically for medical tabular data. This framework consists two critical components: a tabular cell embedding generator and a tabular transformer. The former efficiently encodes diverse modalities from both structured and unstructured tabular data into a harmonized language semantic space with the help of pre-trained sentence encoder and medical prompts. The latter integrates cell representations to generate patient embeddings for various medical tasks. In comprehensive experiments on two real-world datasets for three medical tasks, P-Transformer demonstrated the improvements with 10.9%/11.0% on RMSE/MAE, 0.5%/2.2% on RMSE/MAE, and 1.6%/0.8% on BACC/AUROC compared to state-of-the-art (SOTA) baselines in predictability. Notably, the model exhibited strong resilience to data corruption in the structured data, particularly when the corruption rates are high.
We present a multimodal deep learning (MDL) framework for predicting physical properties of a 10-dimensional acrylic polymer composite material by merging physical attributes and chemical data. Our MDL model comprises four modules, including three generative deep learning models for material structure characterization and a fourth model for property prediction. Our approach handles an 18-dimensional complexity, with 10 compositional inputs and 8 property outputs, successfully predicting 913,680 property data points across 114,210 composition conditions. This level of complexity is unprecedented in computational materials science, particularly for materials with undefined structures. We propose a framework to analyze the high-dimensional information space for inverse material design, demonstrating flexibility and adaptability to various materials and scales, provided sufficient data is available. This study advances future research on different materials and the development of more sophisticated models, drawing us closer to the ultimate goal of predicting all properties of all materials.
We present LLaMA-Adapter, a lightweight adaption method to efficiently fine-tune LLaMA into an instruction-following model. Using 52K self-instruct demonstrations, LLaMA-Adapter only introduces 1.2M learnable parameters upon the frozen LLaMA 7B model, and costs less than one hour for fine-tuning on 8 A100 GPUs. Specifically, we adopt a set of learnable adaption prompts, and prepend them to the word tokens at higher transformer layers. Then, a zero-initialized attention mechanism with zero gating is proposed, which adaptively injects the new instructional cues into LLaMA, while effectively preserves its pre-trained knowledge. With our efficient training, LLaMA-Adapter can generate high-quality responses, comparable to Alpaca with fully fine-tuned 7B parameters. Besides language commands, our approach can be simply extended to multi-modal instructions for learning image-conditioned LLaMA model, which achieves superior reasoning performance on ScienceQA and COCO Caption benchmarks. Furthermore, we also evaluate the zero-initialized attention mechanism for fine-tuning other pre-trained models (ViT, RoBERTa) on traditional vision and language tasks, demonstrating the superior generalization capacity of our approach. Code is released at https://github.com/OpenGVLab/LLaMA-Adapter.
Differentiating tumor progression (TP) from treatment-related necrosis (TN) is critical for clinical management decisions in glioblastoma (GBM). Dynamic FDG PET (dPET), an advance from traditional static FDG PET, may prove advantageous in clinical staging. dPET includes novel methods of a model-corrected blood input function that accounts for partial volume averaging to compute parametric maps that reveal kinetic information. In a preliminary study, a convolution neural network (CNN) was trained to predict classification accuracy between TP and TN for $35$ brain tumors from $26$ subjects in the PET-MR image space. 3D parametric PET Ki (from dPET), traditional static PET standardized uptake values (SUV), and also the brain tumor MR voxels formed the input for the CNN. The average test accuracy across all leave-one-out cross-validation iterations adjusting for class weights was $0.56$ using only the MR, $0.65$ using only the SUV, and $0.71$ using only the Ki voxels. Combining SUV and MR voxels increased the test accuracy to $0.62$. On the other hand, MR and Ki voxels increased the test accuracy to $0.74$. Thus, dPET features alone or with MR features in deep learning models would enhance prediction accuracy in differentiating TP vs TN in GBM.