Recent advances in Neural Architecture Search (NAS) are essentially attributed to Performance Estimation (PE), i.e., a method aims to effectively estimate an architecture. Meanwhile, Kendall's $\tau$τ is well recognized as the principled evaluation criteria for PE strategies in the literature. We argue that Kendall's $\tau$τ is not the optimal solution. Through extensive experiments and theoretical analysis, we take the initiative to reveal the problem behind the Kendall's $\tau$τ and propose a novel criterion named Minimum Keeping Ratio (MKR), which is closely connected to the final performance of NAS. It allows us to compare different PE approaches in a unified perspective, and use effective ablation studies to verify common beliefs and key differences of PE strategies. Based on the findings from MKR, we are able to derive a simple NAS method by integrating different PE strategies with random sampling. Such a method shows very strong performance in efficiency and effectiveness through extensive experiments on different challenging benchmarks. In particular, our simple random sampling NAS finds the optimal architecture in NASbenchMacro, NASbench201, and NASbench301. It is also well generalized to different search spaces (MobileNet) and tasks (semantic segmentation), finding an architecture surpasses the previous state-of-the-art architectures by 4.25 mIoU under $ 600M$600M FLOPs on ADE20K. Codes are available at https://anonymous.4open.science/r/Anonymization11264.

Most prior robot learning methods focus on image-based observations, limiting their capability in 3-D robotic manipulation. Voxel representation naturally delivers rich spatial features but remains underutilized. Specifically, current voxel-based methods struggle with fine-grained tasks, since precise actions are not fully achievable. However, humans can accomplish these tasks well using vision and proprioception. Inspired by this, this article proposed a novel Fusion-Perception-to-Action Transformer (FP2AT) with cross-layer feature aggregation to handle fine-grained manipulation in 3-D space. In particular, a multiscale 3-D visual fusion attention mechanism is devised to draw attention to local regions of interest and maintain awareness of global scenes, thereby boosting the capabilities of visual perception and action planning. Meanwhile, a 3-D visual mutual attention mechanism is designed and it can also enhance spatial perception. Besides, we further explore the potential of FP2AT by developing its coarse-to-fine version, which progressively refines the action space for more precise predictions. In addition, a proprioceptive encoder is developed to mimic the perception of body movements and contact, elevating the effectiveness of the FP2AT. Furthermore, a new metric, the average number of key actions (ANKA), is introduced to evaluate efficiency and planning capability. In various simulated and real-robot examples, our methods significantly outperform state-of-the-art 3-D-vision-based methods in success rate and ANKA metrics.

Existing multi-label knowledge distillation methods simply use regression or single-label classification methods without fully exploiting the essence of multi-label classification, resulting in student models’ inadequate performance and poor functional similarity to teacher models. In this paper, we reinterpret multi-label classification as multiple intra-class ranking tasks, with each class corresponding to a ranking task. Furthermore, we define the knowledge of multi-label classification models as the ranking of intra-class samples. On the one hand, we propose to evaluate the functional similarity between multi-label classification models with Kendall’s tau and rank-biased overlap, which are common metrics for evaluating ranking similarity. On the other hand, we propose a new functionally similar multi-label knowledge distillation method called FSD, which enables student models to learn the ranking of intra-class samples from teacher models. Finally, experimental results validate that FSD outperforms existing methods, especially for functional similarity. Specifically, we achieve a mAP of 73.38% and a mKDT of 0.686 on COCO, which are 2.22% and 0.19 better than existing methods, respectively.