Federated Split Learning (FSL) is a promising distributed learning paradigm in practice, which gathers the strengths of both Federated Learning (FL) and Split Learning (SL) paradigms, to ensure model privacy while diminishing the resource overhead of each client, especially on large transformer models in a resource-constrained environment, e.g., Internet of Things (IoT). However, almost all works merely investigate the performance with simple neural network models in FSL. Despite the minor efforts focusing on incorporating Vision Transformers (ViT) as model architectures, they train ViT from scratch, thereby leading to enormous training overhead in each device with limited resources. Therefore, in this paper, we harness Pre-trained Image Transformers (PITs) as the initial model, coined FES-PIT, to accelerate the training process and improve model robustness. Furthermore, we propose FES-PTZO to hinder the gradient inversion attack, especially having the capability compatible with black-box scenarios, where the gradient information is unavailable. Concretely, FES-PTZO approximates the server gradient by utilizing a zeroth-order (ZO) optimization, which replaces the backward propagation with just one forward process. Empirically, we are the first to provide a systematic evaluation of FSL methods with PITs in real-world datasets, different partial device participations, and heterogeneous data splits. Our experiments verify the effectiveness of our algorithms.
Neural radiance fields~(NeRF) have recently been applied to render large-scale scenes. However, their limited model capacity typically results in blurred rendering results. Existing large-scale NeRFs primarily address this limitation by partitioning the scene into blocks, which are subsequently handled by separate sub-NeRFs. These sub-NeRFs, trained from scratch and processed independently, lead to inconsistencies in geometry and appearance across the scene. Consequently, the rendering quality fails to exhibit significant improvement despite the expansion of model capacity. In this work, we present global-guided focal neural radiance field (GF-NeRF) that achieves high-fidelity rendering of large-scale scenes. Our proposed GF-NeRF utilizes a two-stage (Global and Focal) architecture and a global-guided training strategy. The global stage obtains a continuous representation of the entire scene while the focal stage decomposes the scene into multiple blocks and further processes them with distinct sub-encoders. Leveraging this two-stage architecture, sub-encoders only need fine-tuning based on the global encoder, thus reducing training complexity in the focal stage while maintaining scene-wide consistency. Spatial information and error information from the global stage also benefit the sub-encoders to focus on crucial areas and effectively capture more details of large-scale scenes. Notably, our approach does not rely on any prior knowledge about the target scene, attributing GF-NeRF adaptable to various large-scale scene types, including street-view and aerial-view scenes. We demonstrate that our method achieves high-fidelity, natural rendering results on various types of large-scale datasets. Our project page: https://shaomq2187.github.io/GF-NeRF/
This paper introduces GAgent: an Gripping Agent designed for open-world environments that provides advanced cognitive abilities via VLM agents and flexible grasping abilities with variable stiffness soft grippers. GAgent comprises three primary components - Prompt Engineer module, Visual-Language Model (VLM) core and Workflow module. These three modules enhance gripper success rates by recognizing objects and materials and accurately estimating grasp area even under challenging lighting conditions. As part of creativity, researchers also created a bionic hybrid soft gripper with variable stiffness capable of gripping heavy loads while still gently engaging objects. This intelligent agent, featuring VLM-based cognitive processing with bionic design, shows promise as it could potentially benefit UAVs in various scenarios.
Normalizing flow is a generative modeling approach with efficient sampling. However, Flow-based models suffer two issues, which are manifold and discrete data. If the target distribution is a manifold, which means the dimension of the latent target distribution and the dimension of the data distribution are unmatched, flow-based models might perform badly. Discrete data makes flow-based models collapse into a degenerate mixture of point masses. In this paper, to sidestep such two issues we propose PaddingFlow, a novel dequantization method, which improves normalizing flows with padding-dimensional noise. PaddingFlow is easy to implement, computationally cheap, widely suitable for various tasks, and generates samples that are unbiased estimations of the data. Especially, our method can overcome the limitation of existing dequantization methods that have to change the data distribution, which might degrade performance. We validate our method on the main benchmarks of unconditional density estimation, including five tabular datasets and four image datasets for VAE models, and the IK experiments which are conditional density estimation. The results show that PaddingFlow can provide improvement on all tasks in this paper.
Offline goal-conditioned reinforcement learning (GCRL) aims at solving goal-reaching tasks with sparse rewards from an offline dataset. While prior work has demonstrated various approaches for agents to learn near-optimal policies, these methods encounter limitations when dealing with diverse constraints in complex environments, such as safety constraints. Some of these approaches prioritize goal attainment without considering safety, while others excessively focus on safety at the expense of training efficiency. In this paper, we study the problem of constrained offline GCRL and propose a new method called Recovery-based Supervised Learning (RbSL) to accomplish safety-critical tasks with various goals. To evaluate the method performance, we build a benchmark based on the robot-fetching environment with a randomly positioned obstacle and use expert or random policies to generate an offline dataset. We compare RbSL with three offline GCRL algorithms and one offline safe RL algorithm. As a result, our method outperforms the existing state-of-the-art methods to a large extent. Furthermore, we validate the practicality and effectiveness of RbSL by deploying it on a real Panda manipulator. Code is available at https://github.com/Sunlighted/RbSL.git.
Self-alignment is an effective way to reduce the cost of human annotation while ensuring promising model capability. However, most current methods complete the data collection and training steps in a single round, which may overlook the continuously improving ability of self-aligned models. This gives rise to a key query: What if we do multi-time bootstrapping self-alignment? Does this strategy enhance model performance or lead to rapid degradation? In this paper, our pioneering exploration delves into the impact of bootstrapping self-alignment on large language models. Our findings reveal that bootstrapping self-alignment markedly surpasses the single-round approach, by guaranteeing data diversity from in-context learning. To further exploit the capabilities of bootstrapping, we investigate and adjust the training order of data, which yields improved performance of the model. Drawing on these findings, we propose Step-On-Feet Tuning (SOFT) which leverages model's continuously enhanced few-shot ability to boost zero or one-shot performance. Based on easy-to-hard training recipe, we propose SOFT+ which further boost self-alignment's performance. Our experiments demonstrate the efficiency of SOFT (SOFT+) across various classification and generation tasks, highlighting the potential of bootstrapping self-alignment on continually enhancing model alignment performance.
The classical path planners, such as sampling-based path planners, have the limitations of sensitivity to the initial solution and slow convergence to the optimal solution. However, finding a near-optimal solution in a short period is challenging in many applications such as the autonomous vehicle with limited power/fuel. To achieve an end-to-end near-optimal path planner, we first divide the path planning problem into two subproblems, which are path's space segmentation and waypoints generation in the given path's space. We further propose a two-level cascade neural network named Path Planning Network (PPNet) to solve the path planning problem by solving the abovementioned subproblems. Moreover, we propose a novel efficient data generation method for path planning named EDaGe-PP. The results show the total computation time is less than 1/33 and the success rate of PPNet trained by the dataset that is generated by EDaGe-PP is about $2 \times$ compared to other methods. We validate PPNet against state-of-the-art path planning methods. The results show PPNet can find a near-optimal solution in 15.3ms, which is much shorter than the state-of-the-art path planners.
Replaying past experiences has proven to be a highly effective approach for averting catastrophic forgetting in supervised continual learning. However, some crucial factors are still largely ignored, making it vulnerable to serious failure, when used as a solution to forgetting in continual reinforcement learning, even in the context of perfect memory where all data of previous tasks are accessible in the current task. On the one hand, since most reinforcement learning algorithms are not invariant to the reward scale, the previously well-learned tasks (with high rewards) may appear to be more salient to the current learning process than the current task (with small initial rewards). This causes the agent to concentrate on those salient tasks at the expense of generality on the current task. On the other hand, offline learning on replayed tasks while learning a new task may induce a distributional shift between the dataset and the learned policy on old tasks, resulting in forgetting. In this paper, we introduce RECALL, a replay-enhanced method that greatly improves the plasticity of existing replay-based methods on new tasks while effectively avoiding the recurrence of catastrophic forgetting in continual reinforcement learning. RECALL leverages adaptive normalization on approximate targets and policy distillation on old tasks to enhance generality and stability, respectively. Extensive experiments on the Continual World benchmark show that RECALL performs significantly better than purely perfect memory replay, and achieves comparable or better overall performance against state-of-the-art continual learning methods.
Goal-conditioned rearrangement of deformable objects (e.g. straightening a rope and folding a cloth) is one of the most common deformable manipulation tasks, where the robot needs to rearrange a deformable object into a prescribed goal configuration with only visual observations. These tasks are typically confronted with two main challenges: the high dimensionality of deformable configuration space and the underlying complexity, nonlinearity and uncertainty inherent in deformable dynamics. To address these challenges, we propose a novel representation strategy that can efficiently model the deformable object states with a set of keypoints and their interactions. We further propose local-graph neural network (GNN), a light local GNN learning to jointly model the deformable rearrangement dynamics and infer the optimal manipulation actions (e.g. pick and place) by constructing and updating two dynamic graphs. Both simulated and real experiments have been conducted to demonstrate that the proposed dynamic graph representation shows superior expressiveness in modeling deformable rearrangement dynamics. Our method reaches much higher success rates on a variety of deformable rearrangement tasks (96.3% on average) than state-of-the-art method in simulation experiments. Besides, our method is much more lighter and has a 60% shorter inference time than state-of-the-art methods. We also demonstrate that our method performs well in the multi-task learning scenario and can be transferred to real-world applications with an average success rate of 95% by solely fine tuning a keypoint detector.