UniTAC: Universal Task-Aware Compression via Weighted Distortion Measures
2026-08-17 • Machine Learning
Machine LearningArtificial IntelligenceInformation TheoryMultimedia
AI summaryⓘ
The authors introduce UniTAC, an image compression system that adapts to different tasks without needing retraining. Instead of making a separate codec for each task, UniTAC adjusts by sending a small task-specific signal that tells both encoder and decoder what parts of the image are important. This method keeps one main model and changes its output quality based on the task at hand, balancing general use and task-focused detail. They tested UniTAC on a vision task and found it performs almost as well as specialized codecs and better than universal codecs. Their design is based on understanding how weighting parts of an image can help improve task-specific performance.
Physical AIimage codectask-specific compressionweighted rate-distortiongradient attributionencoder-decoderVision Transformer (ViT)bitrate (bpp)task conditioning
Authors
Homa Esfahanizadeh, Matin Mortaheb, Jinfeng Du, Harish Viswanathan
Abstract
Physical AI systems such as autonomous vehicles and robots rely on timely exchange of high-dimensional sensory signals under tight bandwidth, latency, and energy budgets. Because the task driving downstream decisions evolves over time, a task-specific codec is brittle and retraining one per task is infeasible in the field. We propose UniTAC, a single learned image codec spanning universal (task-agnostic) to task-specialized operation, re-targeted at runtime without retraining. The task is abstracted as a per-component importance vector, derived, e.g., from gradient attribution of any downstream model, and transmitted as low-overhead side information that conditions both encoder and decoder. Trained once over a broad, randomized family of such vectors against weighted-reconstruction distortion, UniTAC keeps a fixed backbone and a single human-viewable reconstruction whose fidelity is steered to the active task by swapping the injected vector. We analyze the underlying weighted rate-distortion problem, characterizing when a diagonal weighted distortion is task-consistent and how weights relate to task sensitivity. Guided by this, we design a Vision Transformer (ViT) codec whose token-level conditioning natively realizes this weight-driven code. On a localized task at 0.034 bpp, a single UniTAC model reaches 91.4% accuracy, only 1.9% below a task-based codec (93.3%) and above universal codecs (76.9%).