Survey of multimodal deep generative models for diverse data types.
problem Inference of shared representations and cross-modal generation from heterogeneous multimodal data.
method Variational autoencoders and other deep generative models.
result A comprehensive survey of multimodal deep generative models.
Plug-and-play multimodal controller improves class-conditional image generation.
problem Generating class-conditional images from user-specified labels.
method Introduces a `multimodal controller` to generate multimodal data without additional learning parameters.
result Multimodal controlled generative models produce higher quality class-conditional images and novel modalities.
Develops a contrastive framework for data-efficient multimodal learning.
problem Expensive training of multimodal generative models requiring related multimodal data.
method Contrastive framework for multimodal learning, distinguishing related from unrelated data.
result Data-efficient multimodal learning on challenging datasets for various VAE models.
Learning multimodal representations is a fundamentally complex research problem due to the presence of multiple heterogeneous sources of information. Although the presence of multiple modalities provides additional valuable information, there are two key challenges to address when learning from multimodal data: 1) mode…
Framework generates multimodal datasets with known MI for benchmarking.
problem Benchmarking mutual information estimators and SSL techniques.
method Flow-based generative model and structured causal framework.
result Regression performance improves with increasing MI between modalities.
Improved text-to-image and multimodal understanding through adaptive generation order optimization.
problem Determining optimal generation sequences in text-to-image synthesis and multimodal understanding.
method Introduced a learnable control module trained via Group Relative Policy Optimization (GRPO) to determine the generation order.
result Learning the control block substantially improves text-to-image alignment and multimodal understanding in DLMs.
New results show contrastive learning can recover shared factors in multimodal data.
problem Understanding when contrastive learning can recover shared latent factors in multimodal data.
method New identifiability results for multimodal contrastive learning, distinguishing between multi-view and multimodal settings.
result Contrastive learning can block-identify shared latent factors in multimodal data, even with dependencies.
New ELBO formulation improves multimodal learning.
problem Existing ELBO models struggle with multimodal data.
method Proposes a generalized ELBO for multimodal data.
result Demonstrates improved performance over state-of-the-art models.
Unified approach for multimodal data prediction using synthetic data generation.
problem Challenges in integrating heterogeneous data types for accurate predictive performance.
method Generative Distribution Prediction (GDP) framework that uses multimodal synthetic data generation.
result Empirical validation across four tasks demonstrates versatility and effectiveness of GDP.
The paper introduces multimodal generative models to improve data marginal likelihood.
problem Improving data marginal likelihood in multimodal settings.
method Derives variational bounds on the evidence for multimodal deep generative models, generalizes objectives for different model types, and benchmarks across various datasets.
result Multimodal VAEs excel in image, label, and text datasets with and without weak supervision.
A new objective function using Jensen-Shannon divergence improves generative learning from multiple data types.
problem Learning from multiple data types efficiently and accurately.
method Proposes a novel objective function using Jensen-Shannon divergence to approximate multimodal posteriors directly.
result The mmJSD objective optimizes an ELBO and improves generative learning tasks.
Integrating visual and linguistic information into a single multimodal representation is an unsolved problem with wide-reaching applications to both natural language processing and computer vision. In this paper, we present a simple method to build multimodal representations by learning a language-to-vision mapping and…
DFMs enable flow-based models for multimodal discrete and continuous data.
problem Combining discrete and continuous data for generative models.
method Discrete Flow Models (DFMs) using Continuous Time Markov Chains.
result DFMs achieve state-of-the-art co-design performance for protein structure and sequence generation.
Paper proposes multimodal contrastive learning for EHR data.
problem Separate treatment of structured and unstructured EHR data.
method Proposes a multimodal feature embedding generative model and a multimodal contrastive loss.
result Multimodal learning yields better feature representation than single-modality learning.
VAEs struggle with surjective multimodal data, especially class labels describing images.
problem VAEs struggle to capture variability in surjective multimodal data.
method Theoretical and empirical demonstration of VAEs with a mixture of experts posterior.
result VAEs with a mixture of experts posterior can disregard variation in surjective multimodal data.
Generative Score Inference improves uncertainty quantification for multimodal data.
problem Accurate uncertainty quantification in multimodal learning tasks.
method Generative Score Inference (GSI) uses synthetic samples to approximate conditional score distributions.
result GSI achieves state-of-the-art performance in hallucination detection and image captioning uncertainty estimation.
Improved multimodal variational models capture more complex joint distributions.
problem Limited expressiveness of multimodal variational models.
method Used normalizing flows to approximate and transform a simple parametric joint posterior into a more complex one.
result The model improves on state-of-the-art multimodal variational methods on various tasks.
The complex world around us is inherently multimodal and sequential (continuous). Information is scattered across different modalities and requires multiple continuous sensors to be captured. As machine learning leaps towards better generalization to real world, multimodal sequential learning becomes a fundamental rese…
Evidential Softmax preserves multimodality in sparse probability distributions for generative models.
problem Sparse probability distributions in deep generative models make exact marginalization computationally intractable.
method Introduce ev-softmax, a sparse normalization function that preserves multimodality and can be trained with probabilistic loss functions.
result ev-softmax outperforms existing techniques in distributional accuracy and dimensionality reduction.
Generates multimodal safety-critical scenarios for robustness evaluation of decision-making algorithms.
problem Lack of comprehensive evaluation of neural network robustness under real-world scenarios.
method Proposes a flow-based multimodal scenario generator using weighted likelihood maximization and gradient-based sampling.
result Demonstrates improved testing efficiency and multimodal modeling capability compared to traditional methods.
Hybrid model for multimodal distributions using diffusion and classification.
problem Sampling from multimodal distributions with correct proportions.
method Divide-and-conquer strategy: identify modes, train classifiers, diffusion models, bridge sampling.
result Framework effectively handles multimodal distributions in high dimensions.
Estimates interactions between modalities for multimodal data.
problem Accurately quantifying interactions between different data types.
method Developed Lightweight Sample-wise Multimodal Interaction (LSMI) estimator using pointwise information theory.
result LSMI reveals fine-grained dynamics in multimodal data.
Enhances multimodal generation with Normalizing Flows and correlation analysis.
problem Generating coherent cross-modal data from multiple sources.
method Uses Deep Canonical Correlation Analysis for shared information, Normalizing Flows for diversity, and Product of Experts for scalability.
result Improves likelihood, diversity, and coherence in conditional generation.
Multimodal machine learning is a core research area spanning the language, visual and acoustic modalities. The central challenge in multimodal learning involves learning representations that can process and relate information from multiple modalities. In this paper, we propose two methods for unsupervised learning of j…
In this paper we study how to learn stochastic, multimodal transition dynamics in reinforcement learning (RL) tasks. We focus on evaluating transition function estimation, while we defer planning over this model to future work. Stochasticity is a fundamental property of many task environments. However, discriminative f…
Generative Stochastic Networks (GSNs) have been recently introduced as an alternative to traditional probabilistic modeling: instead of parametrizing the data distribution directly, one parametrizes a transition operator for a Markov chain whose stationary distribution is an estimator of the data generating distributio…
MHVAE learns cross-modality inference inspired by human cognition.
problem Cross-modality inference in multimodal data.
method Hierarchical multimodal generative model with modality-specific and joint-modality distributions.
result MHVAE performs on par with state-of-the-art models on multimodal datasets.
SGLB boosts machine learning with Langevin diffusion for multimodal loss functions.
problem Dealing with multimodal loss functions in machine learning.
method Stochastic Gradient Langevin Boosting (SGLB) based on Langevin diffusion equation.
result SGLB guarantees global convergence for multimodal loss functions.
GGMPs improve non-Gaussian conditional density estimation.
problem Multimodality, heteroscedasticity, and strong non-Gaussianity in conditional density estimation.
method GGMP combines local Gaussian mixture fitting, cross-input component alignment, and per-component heteroscedastic GP training.
result GGMPs improve distributional approximation on synthetic and real-world datasets.
Open-FinLLMs tackle financial tasks with multimodal capabilities.
problem Financial LLMs lack multimodal capabilities and real-world applicability.
method Developed Open-FinLLMs, an open-source multimodal financial LLM suite.
result Open-FinLLMs outperform advanced financial and general LLMs in diverse tasks.
MDNs offer a data-efficient alternative to diffusion and flow models for multimodal scientific learning.
problem Capturing multimodal conditional uncertainty in scientific inverse problems.
method Mixture Density Networks (MDNs) as explicit parametric density estimators.
result MDNs achieve superior generalization, interpretability, and sample efficiency in scientific tasks.
Multimodal fusion is considered a key step in multimodal tasks such as sentiment analysis, emotion detection, question answering, and others. Most of the recent work on multimodal fusion does not guarantee the fidelity of the multimodal representation with respect to the unimodal representations. In this paper, we prop…
Multimodal learning aims to discover the relationship between multiple modalities. It has become an important research topic due to extensive multimodal applications such as cross-modal retrieval. This paper attempts to address the modality heterogeneity problem based on Gaussian process latent variable models (GPLVMs)…
Dictionary learning algorithms have been successfully used for both reconstructive and discriminative tasks, where an input signal is represented with a sparse linear combination of dictionary atoms. While these methods are mostly developed for single-modality scenarios, recent studies have demonstrated the advantages …
Multimodal research is an emerging field of artificial intelligence, and one of the main research problems in this field is multimodal fusion. The fusion of multimodal data is the process of integrating multiple unimodal representations into one compact multimodal representation. Previous research in this field has exp…
Push-forward models struggle to fit multimodal distributions due to high Lipschitz constants.
problem Expressivity of push-forward generative models in fitting multimodal distributions.
method Analyzing the Lipschitz constant and its relation to the total variation distance and Kullback-Leibler divergence.
result Push-forward models require high Lipschitz constants to approximate multimodal distributions, leading to a trade-off between expressivity and stability.
New model learns multimodal data better than DAGs.
problem Complex multimodal data not well captured by DAGs.
method Latent partial causal model with two latent coupled variables.
result Identifiability result shows representations correspond to latent variables.
Proposes a multimodal deep generative model for semi-supervised learning with class imbalance.
problem Class imbalance in semi-supervised learning with partial supervision.
method Separate encoders for each modality, sharing latent variables, and using Student's t-distributions for prior, encoder, and decoder.
result Outperforms baseline methods in generalization and classification performance for partially labeled multimodal data.
Improved sentiment analysis with multimodal data.
problem Cross-modal sentiment analysis in social media, customer service, and video blogs.
method Gated mechanism for attention-based learning of cross-modal interactions, with experiments on CMU-MOSI and CMU-MOSEI datasets.
result 1.6% and 1.34% absolute improvement over state-of-the-art.
Multimodalities provide promising performance than unimodality in most tasks. However, learning the semantic of the representations from multimodalities efficiently is extremely challenging. To tackle this, we propose the Transformer based Cross-modal Translator (TCT) to learn unimodal sequence representations by trans…
Humor is a unique and creative communicative behavior displayed during social interactions. It is produced in a multimodal manner, through the usage of words (text), gestures (vision) and prosodic cues (acoustic). Understanding humor from these three modalities falls within boundaries of multimodal language; a recent r…
New framework estimates graph from multimodal functional data.
problem Estimating graph from joint multimodal functional data.
method Integrative framework using partial correlation operator.
result Estimator converges to stationary point with quantifiable error.
Representing entities and relations in an embedding space is a well-studied approach for machine learning on relational data. Existing approaches, however, primarily focus on simple link structure between a finite set of entities, ignoring the variety of data types that are often used in knowledge bases, such as text, …
Systematic review of multimodal data challenges and solutions.
problem Challenges in integrating diverse data types for improved diagnostics and personalized care.
method Synthesizing findings from 69 studies on technical obstacles and recent methodological advances.
result Promising solutions like transfer learning, generative models, attention mechanisms, and neural architecture search.
Extends SGM to functional spaces for multimodal data.
problem Modeling densities in functional spaces.
method Represent data in spectral space, dissociate stochastic and space-time components, use SGM for sampling.
result Demonstrates effectiveness on multimodal datasets.
A new method improves Bayesian inference for multimodal posteriors.
problem Insensitivity to well-separated modes in multimodal posteriors.
method Weighted Kernel Stein Discrepancy method.
result Significantly improved mode sensitivity compared to standard KSD-Bayes.
We study the problem of multimodal generative modelling of images based on generative adversarial networks (GANs). Despite the success of existing methods, they often ignore the underlying structure of vision data or its multimodal generation characteristics. To address this problem, we introduce the Dirichlet prior fo…
Computational modeling of human multimodal language is an emerging research area in natural language processing spanning the language, visual and acoustic modalities. Comprehending multimodal language requires modeling not only the interactions within each modality (intra-modal interactions) but more importantly the in…