{"ID":2868004,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.16853","arxiv_id":"2509.16853","title":"ISCS: Parameter-Guided Feature Pruning for Resource-Constrained Embodied Perception","abstract":"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.","short_abstract":"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 representat...","url_abs":"https://arxiv.org/abs/2509.16853","url_pdf":"https://arxiv.org/pdf/2509.16853v2","authors":"[\"Jinhao Wang\",\"Nam Ling\",\"Wei Wang\",\"Wei Jiang\"]","published":"2025-09-21T00:44:15Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Generative Adversarial Network\"]","has_code":false}