OmniVec2 -- A Novel Transformer based Network for Large Scale Multimodal and Multitask Learning

TL;DR

OmniVec2 introduces a transformer-based network capable of handling 12 diverse data modalities for large-scale multimodal and multitask learning, achieving state-of-the-art results through innovative training strategies.

Contribution

It presents a new multimodal multitask network with modality-specific tokenizers, a shared transformer, and a novel pretraining and training algorithm for diverse data modalities.

Findings

Achieves state-of-the-art performance across 25 datasets from 12 modalities.

Effectively unifies multiple modalities into a shared embedding space.

Demonstrates robustness and versatility in multimodal multitask learning.

Abstract

We present a novel multimodal multitask network and associated training algorithm. The method is capable of ingesting data from approximately 12 different modalities namely image, video, audio, text, depth, point cloud, time series, tabular, graph, X-ray, infrared, IMU, and hyperspectral. The proposed approach utilizes modality specialized tokenizers, a shared transformer architecture, and cross-attention mechanisms to project the data from different modalities into a unified embedding space. It addresses multimodal and multitask scenarios by incorporating modality-specific task heads for different tasks in respective modalities. We propose a novel pretraining strategy with iterative modality switching to initialize the network, and a training algorithm which trades off fully joint training over all modalities, with training on pairs of modalities at a time. We provide comprehensive evaluation across 25 datasets from 12 modalities and show state of the art performances, demonstrating the effectiveness of the proposed architecture, pretraining strategy and adapted multitask training.