# ISCS: Parameter-Guided Feature Pruning for Resource-Constrained Embodied Perception

arXiv: https://arxiv.org/abs/2509.16853 · HTML: https://mlsanity.com/paper/2509.16853/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 12 references that are in the map.
- First posted on arXiv 21 Sep 2025.

## Builds on (most cited first, 9 of 9)

- [Scalable Image Coding for Humans and Machines](https://mlsanity.com/paper/2107.08373/) (2021)
- [MLIC++: Linear Complexity Multi-Reference Entropy Modeling for Learned Image Compression](https://mlsanity.com/paper/2307.15421/) (2023)
- [DC-AE 1.5: Accelerating Diffusion Model Convergence with Structured Latent Space](https://mlsanity.com/paper/2508.00413/) (2025)
- [Image Compression for Machine and Human Vision with Spatial-Frequency Adaptation](https://mlsanity.com/paper/2407.09853/) (2024)
- [Task-Oriented Feature Compression for Multimodal Understanding via Device-Edge Co-Inference](https://mlsanity.com/paper/2503.12926/) (2025)
- [Energy-Efficient Edge Inference in Integrated Sensing, Communication, and Computation Networks](https://mlsanity.com/paper/2503.00298/) (2025)
- [DeepFGS: Fine-Grained Scalable Coding for Learned Image Compression](https://mlsanity.com/paper/2412.00437/) (2024)
- [Latency-Memory Optimized Splitting of Convolution Neural Networks for Resource Constrained Edge Devices](https://mlsanity.com/paper/2107.09123/) (2021)
- [EdgeMM: Multi-Core CPU with Heterogeneous AI-Extension and Activation-aware Weight Pruning for Multimodal LLMs at Edge](https://mlsanity.com/paper/2505.10782/) (2025)

## Abstract (arXiv)

Prior studies in embodied AI consistently show that robust perception is critical for human-robot interaction, yet deploying high-fidelity visual models on resource-constrained agents remains challenging due to limited on-device computation power and transmission latency. Exploiting the redundancy in latent representations could improve system efficiency, yet existing approaches often rely on costly dataset-specific ablation tests or heavy entropy models unsuitable for real-time edge-robot collaboration. We propose a generalizable, dataset-agnostic method to identify and selectively transmit structure-critical channels in pretrained encoders. Instead of brute-force empirical evaluations, our approach leverages intrinsic parameter statistics-weight variances and biases-to estimate channel importance. This analysis reveals a consistent organizational structure, termed the Invariant Salient Channel Space (ISCS), where Salient-Core channels capture dominant structures while Salient-Auxiliary channels encode fine visual details. Building on ISCS, we introduce a deterministic static pruning strategy that enables lightweight split-computing. Experiments across different datasets demonstrate that our method achieves a deterministic, ultra-low latency pipeline by bypassing heavy entropy modeling. Our method reduces end-to-end latency, providing a critical speed-accuracy trade-off for resource-constrained human-aware embodied systems.