Existing methods for measuring training sample influence on models only provide static, overall measurements, overlooking how sample influence changes during training. We propose Dynamic Influence Tracker (DIT), which captures the time-varying sample influence across arbitrary time windows during training. DIT offers three key insights: 1) Samples show different time-varying influence patterns, with some samples important in the early training stage while others become important later. 2) Sample influences show a weak correlation between early and late stages, demonstrating that the model undergoes distinct learning phases with shifting priorities. 3) Analyzing influence during the convergence period provides more efficient and accurate detection of corrupted samples than full-training analysis. Supported by theoretical guarantees without assuming loss convexity or model convergence, DIT significantly outperforms existing methods, achieving up to 0.99 correlation with ground truth and above 98\% accuracy in detecting corrupted samples in complex architectures.

Time-aware data valuation enhances training efficiency and model robustness, as early detection of harmful samples could prevent months of wasted computation. However, existing methods rely on model retraining or convergence assumptions or fail to capture long-term training dynamics. We propose LiveVal, an efficient time-aware data valuation method with three key designs: 1) seamless integration with SGD training for efficient data contribution monitoring; 2) reference-based valuation with normalization for reliable benchmark establishment; and 3) adaptive reference point selection for real-time updating with optimized memory usage. We establish theoretical guarantees for LiveVal's stability and prove that its valuations are bounded and directionally aligned with optimization progress. Extensive experiments demonstrate that LiveVal provides efficient data valuation across different modalities and model scales, achieving 180 speedup over traditional methods while maintaining robust detection performance.

Although LLMs have achieved significant success, their reliance on large volumes of human-annotated data has limited their potential for further scaling. In this situation, utilizing self-generated synthetic data has become crucial for fine-tuning LLMs without extensive human annotation. However, current methods often fail to ensure consistent improvements across iterations, with performance stagnating after only minimal updates. To overcome these challenges, we introduce Dynamic Noise Preference Optimization (DNPO). DNPO employs a dynamic sample labeling mechanism to construct preference pairs for training and introduces controlled, trainable noise into the preference optimization process. Our approach effectively prevents stagnation and enables continuous improvement. In experiments with Zephyr-7B, DNPO consistently outperforms existing methods, showing an average performance boost of 2.6% across multiple benchmarks. Additionally, DNPO shows a significant improvement in model-generated data quality, with a 29.4% win-loss rate gap compared to the baseline in GPT-4 evaluations. This highlights its effectiveness in enhancing model performance through iterative refinement.

This paper studies an integrated sensing and communications (ISAC) system for low-altitude economy (LAE), where a ground base station (GBS) provides communication and navigation services for authorized unmanned aerial vehicles (UAVs), while sensing the low-altitude airspace to monitor the unauthorized mobile target. The expected communication sum-rate over a given flight period is maximized by jointly optimizing the beamforming at the GBS and UAVs' trajectories, subject to the constraints on the average signal-to-noise ratio requirement for sensing, the flight mission and collision avoidance of UAVs, as well as the maximum transmit power at the GBS. Typically, this is a sequential decision-making problem with the given flight mission. Thus, we transform it to a specific Markov decision process (MDP) model called episode task. Based on this modeling, we propose a novel LAE-oriented ISAC scheme, referred to as Deep LAE-ISAC (DeepLSC), by leveraging the deep reinforcement learning (DRL) technique. In DeepLSC, a reward function and a new action selection policy termed constrained noise-exploration policy are judiciously designed to fulfill various constraints. To enable efficient learning in episode tasks, we develop a hierarchical experience replay mechanism, where the gist is to employ all experiences generated within each episode to jointly train the neural network. Besides, to enhance the convergence speed of DeepLSC, a symmetric experience augmentation mechanism, which simultaneously permutes the indexes of all variables to enrich available experience sets, is proposed. Simulation results demonstrate that compared with benchmarks, DeepLSC yields a higher sum-rate while meeting the preset constraints, achieves faster convergence, and is more robust against different settings.

Research trends in SLAM systems are now focusing more on multi-sensor fusion to handle challenging and degenerative environments. However, most existing multi-sensor fusion SLAM methods mainly use all of the data from a range of sensors, a strategy we refer to as the all-in method. This method, while merging the benefits of different sensors, also brings in their weaknesses, lowering the robustness and accuracy and leading to high computational demands. To address this, we propose a new fusion approach -- Selective Kalman Filter -- to carefully choose and fuse information from multiple sensors (using LiDAR and visual observations as examples in this paper). For deciding when to fuse data, we implement degeneracy detection in LiDAR SLAM, incorporating visual measurements only when LiDAR SLAM exhibits degeneracy. Regarding degeneracy detection, we propose an elegant yet straightforward approach to determine the degeneracy of LiDAR SLAM and to identify the specific degenerative direction. This method fully considers the coupled relationship between rotational and translational constraints. In terms of how to fuse data, we use visual measurements only to update the specific degenerative states. As a result, our proposed method improves upon the all-in method by greatly enhancing real-time performance due to less processing visual data, and it introduces fewer errors from visual measurements. Experiments demonstrate that our method for degeneracy detection and fusion, in addressing degeneracy issues, exhibits higher precision and robustness compared to other state-of-the-art methods, and offers enhanced real-time performance relative to the all-in method. The code is openly available.

Ultra-wideband (UWB) is gaining popularity with devices like AirTags for precise home item localization but faces significant challenges when scaled to large environments like seaports. The main challenges are calibration and localization in obstructed conditions, which are common in logistics environments. Traditional calibration methods, dependent on line-of-sight (LoS), are slow, costly, and unreliable in seaports and warehouses, making large-scale localization a significant pain point in the industry. To overcome these challenges, we propose a UWB-LiDAR fusion-based calibration and one-shot localization framework. Our method uses Gaussian Processes to estimate anchor position from continuous-time LiDAR Inertial Odometry with sampled UWB ranges. This approach ensures accurate and reliable calibration with just one round of sampling in large-scale areas, I.e., 600x450 square meter. With the LoS issues, UWB-only localization can be problematic, even when anchor positions are known. We demonstrate that by applying a UWB-range filter, the search range for LiDAR loop closure descriptors is significantly reduced, improving both accuracy and speed. This concept can be applied to other loop closure detection methods, enabling cost-effective localization in large-scale warehouses and seaports. It significantly improves precision in challenging environments where UWB-only and LiDAR-Inertial methods fall short, as shown in the video \url{https://youtu.be/oY8jQKdM7lU }. We will open-source our datasets and calibration codes for community use.

Existing pruning techniques for large language models (LLMs) targeting domain-specific applications typically follow a two-stage process: pruning the pretrained general-purpose LLMs and then fine-tuning the pruned LLMs on specific domains. However, the pruning decisions, derived from the pretrained weights, remain unchanged during fine-tuning, even if the weights have been updated. Therefore, such a combination of the pruning decisions and the finetuned weights may be suboptimal, leading to non-negligible performance degradation. To address these limitations, we propose ATP: All-in-One Tuning and Structural Pruning, a unified one-stage structural pruning and fine-tuning approach that dynamically identifies the current optimal substructure throughout the fine-tuning phase via a trainable pruning decision generator. Moreover, given the limited available data for domain-specific applications, Low-Rank Adaptation (LoRA) becomes a common technique to fine-tune the LLMs. In ATP, we introduce LoRA-aware forward and sparsity regularization to ensure that the substructures corresponding to the learned pruning decisions can be directly removed after the ATP process. ATP outperforms the state-of-the-art two-stage pruning methods on tasks in the legal and healthcare domains. More specifically, ATP recovers up to 88% and 91% performance of the dense model when pruning 40% parameters of LLaMA2-7B and LLaMA3-8B models, respectively.

Foundation models (FMs) achieve strong performance across diverse tasks with task-specific fine-tuning, yet full parameter fine-tuning is often computationally prohibitive for large models. Parameter-efficient fine-tuning (PEFT) methods like Low-Rank Adaptation (LoRA) reduce this cost by introducing low-rank matrices for tuning fewer parameters. While LoRA allows for efficient fine-tuning, it requires significant data for adaptation, making Federated Learning (FL) an appealing solution due to its privacy-preserving collaborative framework. However, combining LoRA with FL introduces two key challenges: the \textbf{Server-Side LoRA Aggregation Bias}, where server-side averaging of LoRA matrices diverges from the ideal global update, and the \textbf{Client-Side LoRA Initialization Drift}, emphasizing the need for consistent initialization across rounds. Existing approaches address these challenges individually, limiting their effectiveness. We propose LoRA-FAIR, a novel method that tackles both issues by introducing a correction term on the server while keeping the original LoRA modules, enhancing aggregation efficiency and accuracy. LoRA-FAIR maintains computational and communication efficiency, yielding superior performance over state-of-the-art methods. Experimental results on ViT and MLP-Mixer models across large-scale datasets demonstrate that LoRA-FAIR consistently achieves performance improvements in FL settings.

We introduce SPOT, an object-centric imitation learning framework. The key idea is to capture each task by an object-centric representation, specifically the SE(3) object pose trajectory relative to the target. This approach decouples embodiment actions from sensory inputs, facilitating learning from various demonstration types, including both action-based and action-less human hand demonstrations, as well as cross-embodiment generalization. Additionally, object pose trajectories inherently capture planning constraints from demonstrations without the need for manually crafted rules. To guide the robot in executing the task, the object trajectory is used to condition a diffusion policy. We show improvement compared to prior work on RLBench simulated tasks. In real-world evaluation, using only eight demonstrations shot on an iPhone, our approach completed all tasks while fully complying with task constraints. Project page: https://nvlabs.github.io/object_centric_diffusion

Recently, federated multi-view clustering (FedMVC) has emerged to explore cluster structures in multi-view data distributed on multiple clients. Existing approaches often assume that clients are isomorphic and all of them belong to either single-view clients or multi-view clients. Despite their success, these methods also present limitations when dealing with practical FedMVC scenarios involving heterogeneous hybrid views, where a mixture of both single-view and multi-view clients exhibit varying degrees of heterogeneity. In this paper, we propose a novel FedMVC framework, which concurrently addresses two challenges associated with heterogeneous hybrid views, i.e., client gap and view gap. To address the client gap, we design a local-synergistic contrastive learning approach that helps single-view clients and multi-view clients achieve consistency for mitigating heterogeneity among all clients. To address the view gap, we develop a global-specific weighting aggregation method, which encourages global models to learn complementary features from hybrid views. The interplay between local-synergistic contrastive learning and global-specific weighting aggregation mutually enhances the exploration of the data cluster structures distributed on multiple clients. Theoretical analysis and extensive experiments demonstrate that our method can handle the heterogeneous hybrid views in FedMVC and outperforms state-of-the-art methods.