CURA: Size Isnt All You Need - A Compact Universal Architecture for On-Device Intelligence

2025-09-29

AI Summary

Current on-device AI architectures face two key bottlenecks: (1) model parameters scale linearly with task complexity, and (2) poor cross-domain generalization—e.g., regression models cannot handle NLP tasks. To address these, we propose CURA, a compact, universal architecture inspired by analog audio circuits, enabling unified modeling across regression, classification, NLP, and computer vision on resource-constrained devices. Its core innovations include: (i) a network structure grounded in analog signal processing principles; (ii) lightweight parameter-sharing modules; and (iii) dynamic feature modulation for enhanced expressivity and task adaptability under extreme parameter constraints. Experiments demonstrate that CURA matches the performance of domain-specific models (up to 90% F1 score) while reducing parameter counts by up to 2500× and lowering prediction error by 1.6–2.1× across multiple benchmarks.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Computer Vision: Learning & Optimization for CV
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Existing on-device AI architectures for resource-constrained environments face two critical limitations: they lack compactness, with parameter requirements scaling proportionally to task complexity, and they exhibit poor generalizability, performing effectively only on specific application domains (e.g., models designed for regression tasks cannot adapt to natural language processing (NLP) applications). In this paper, we propose CURA, an architecture inspired by analog audio signal processing circuits that provides a compact and lightweight solution for diverse machine learning tasks across multiple domains. Our architecture offers three key advantages over existing approaches: (1) Compactness: it requires significantly fewer parameters regardless of task complexity; (2) Generalizability: it adapts seamlessly across regression, classification, complex NLP, and computer vision tasks; and (3) Complex pattern recognition: it can capture intricate data patterns while maintaining extremely low model complexity. We evaluated CURA across diverse datasets and domains. For compactness, it achieved equivalent accuracy using up to 2,500 times fewer parameters compared to baseline models. For generalizability, it demonstrated consistent performance across four NLP benchmarks and one computer vision dataset, nearly matching specialized existing models (achieving F1-scores up to 90%). Lastly, it delivers superior forecasting accuracy for complex patterns, achieving 1.6 times lower mean absolute error and 2.1 times lower mean squared error than competing models.

Problem

• Addressing lack of compactness in on-device AI architectures
• Overcoming poor generalizability across diverse machine learning tasks
• Enabling complex pattern recognition with minimal model complexity

Innovation

• Compact architecture inspired by analog circuits
• Universal adaptability across diverse AI tasks
• Complex pattern recognition with minimal parameters