Federated unlearning (FU) aims to remove a participant’s data contributions from a trained federated learning (FL) model, ensuring privacy and regulatory compliance. Traditional FU methods often depend on auxiliary storage on either the client or server side or require direct access to the data targeted for removal—a dependency that may not be feasible if the data is no longer available. To overcome these limitations, we propose NoT, a novel and efficient FU algorithm based on weight negation (multiplying by -1), which circumvents the need for additional storage and access to the target data. We argue that effective and efficient unlearning can be achieved by perturbing model parameters away from the set of optimal parameters, yet being well-positioned for quick re-optimization. This technique, though seemingly contradictory, is theoretically grounded: we prove that the weight negation perturbation effectively disrupts inter-layer co-adaptation, inducing unlearning while preserving an approximate optimality property, thereby enabling rapid recovery. Experimental results across three datasets and three model architectures demonstrate that NoT significantly outperforms existing baselines in unlearning efficacy as well as in communication and computational efficiency.

Diffusion models, such as diffusion policy, have achieved state-of-the-art results in robotic manipulation by imitating expert demonstrations. While diffusion models were originally developed for vision tasks like image and video generation, many of their inference strategies have been directly transferred to control domains without adaptation. In this work, we show that by tailoring the denoising process to the specific characteristics of embodied AI tasks—particularly the structured, low-dimensional nature of action distributions---diffusion policies can operate effectively with as few as 5 neural function evaluations (NFE). Building on this insight, we propose a population-based sampling strategy, genetic denoising, which enhances both performance and stability by selecting denoising trajectories with low out-of-distribution risk. Our method solves challenging tasks with only 2 NFE while improving or matching performance. We evaluate our approach across 14 robotic manipulation tasks from D4RL and Robomimic, spanning multiple action horizons and inference budgets. In over 2 million evaluations, our method consistently outperforms standard diffusion-based policies, achieving up to 20\% performance gains with significantly fewer inference steps.

Machine unlearning (MU) aims to remove the influence of specific ''forget'' data from a trained model while preserving its knowledge of the remaining ''retain'' data. Existing MU methods based on label manipulation or model weight perturbations often achieve limited unlearning effectiveness. To address this, we introduce CoUn, a novel MU framework inspired by the observation that a model retrained from scratch using only retain data classifies forget data based on their semantic similarity to the retain data. CoUn emulates this behavior by adjusting learned data representations through contrastive learning (CL) and supervised learning, applied exclusively to retain data. Specifically, CoUn (1) leverages semantic similarity between data samples to indirectly adjust forget representations using CL, and (2) maintains retain representations within their respective clusters through supervised learning. Extensive experiments across various datasets and model architectures show that CoUn consistently outperforms state-of-the-art MU baselines in unlearning effectiveness. Additionally, integrating our CL module into existing baselines empowers their unlearning effectiveness.

Autonomous driving is strongly contingent on perception and motion prediction for scene understanding. In this paper, we propose LIDAR Network (LidNet) to boost perception and motion prediction accuracy by redesigning MotionNet architecture. MotionNet is a new real-time encoder-decoder model that achieves joint perception and motion prediction at a pixel level. LidNet improves MotionNet performance by replacing every two spatial convolution layers in its encoder-decoder architecture with residual blocks and relies on average pooling rather than strided convolution for spatial reduction. In addition, we adjust the lateral skip connections linking encoders and decoders to result in a symmetric network. The global temporal maximum pooling layers on the lateral connections are replaced with temporal average pooling. Further, we introduce a center layer between the encoder-decoder architecture, with no spatial reduction applied at the lowest levels. Our extensive evaluation performed on the nuScenes dataset confirms that LidNet outperforms the state-of-the-art and operates in real-time.

The field of self-driving vehicles (SDVs) is going viral among researchers from a broad spectrum of specialties. SDVs are expected to have profound impacts on the world once fully developed and deployed on roads. Hence, researchers are working assiduously together to accomplish this project. In this paper, we propose integrating motion prediction with sequential latent maximum entropy reinforcement learning, end-to-end, to train an agent to navigate autonomously in a simulated urban environment. The fusion of motion prediction for surrounding vehicles enhances traffic efficiency and safety. A novel network specialized in joint perception and motion prediction, named MotionNet, is selected in our paper to supply us with motion predictions. Our proposed system demonstrates that adding motion prediction enhances performance even further. Furthermore, our system relies merely on LIDAR sensor. CARLA simulator is used to conduct our experiments and extract outcomes.

Perception and motion prediction are indispensable components to the smooth operation of autonomous vehicles and the safety of the surrounding environment. Strengthening the accuracy of perception and motion prediction in autonomous vehicles remains of paramount importance. Therefore, we propose an end-to-end multi-view fusion methodology applied to MotionNet backbone network to enhance the sharpness of both perception and motion prediction. MotionNet is a state-of-the-art real-time model designed for joint perception and motion prediction. Our multi-view input is based on a single LIDAR sensor and formed by the fusion of range view features with bird's eye view. We evaluate our proposed work on nuScenes dataset and demonstrate through experiments that our proposed extension to MotionNet using the multi-view fusion technique outperforms MotionNet in both perception and motion prediction, especially for small and distant objects.

Summary Whale Optimization Algorithm (WOA) is a recent swarm intelligence based meta‐heuristic optimization algorithm, which simulates the natural behavior of bubble‐net hunting strategy of humpback whales and has been successfully applied to solve complex optimization problems in a wide range of disciplines. However, when applied to large‐size problems, its performance degrades due to the need of massive computational work load. Distributed computing is one of the effective ways to improve the scalability of WOA for solving large‐scale problems. In this paper, we propose a simple and robust distributed implementation of WOA using Hadoop MapReduce named MR‐WOA. MapReduce paradigm is adopted as the distribution model since it is one of the most mature technologies to develop parallel algorithms which automatically handles communication, load balancing, data locality, and fault tolerance. The design of MR‐WOA is discussed in details using the MapReduce paradigm. Experiments are conducted for a set of well‐known benchmarks for evaluating the quality, speedup, and scalability of MR‐WOA. The conducted experiments reveal that our approach achieves a promising speedup. For some benchmarks, speedup scales linearly with increasing the number of computational nodes.