To address the growing demand for privacy protection in machine learning, we propose a novel and efficient machine unlearning approach for \textbf{L}arge \textbf{M}odels, called \textbf{LM}Eraser. Existing unlearning research suffers from entangled training data and complex model architectures, incurring extremely high computational costs for large models. LMEraser takes a divide-and-conquer strategy with a prompt tuning architecture to isolate data influence. The training dataset is partitioned into public and private datasets. Public data are used to train the backbone of the model. Private data are adaptively clustered based on their diversity, and each cluster is used to optimize a prompt separately. This adaptive prompt tuning mechanism reduces unlearning costs and maintains model performance. Experiments demonstrate that LMEraser achieves a $100$-fold reduction in unlearning costs without compromising accuracy compared to prior work. Our code is available at: \url{https://github.com/lmeraser/lmeraser}.
Encouraged by the remarkable achievements of language and vision foundation models, developing generalist robotic agents through imitation learning, using large demonstration datasets, has become a prominent area of interest in robot learning. The efficacy of imitation learning is heavily reliant on the quantity and quality of the demonstration datasets. In this study, we aim to scale up demonstrations in a data-efficient way to facilitate the learning of generalist robotic agents. We introduce AdaDemo (Adaptive Online Demonstration Expansion), a general framework designed to improve multi-task policy learning by actively and continually expanding the demonstration dataset. AdaDemo strategically collects new demonstrations to address the identified weakness in the existing policy, ensuring data efficiency is maximized. Through a comprehensive evaluation on a total of 22 tasks across two robotic manipulation benchmarks (RLBench and Adroit), we demonstrate AdaDemo's capability to progressively improve policy performance by guiding the generation of high-quality demonstration datasets in a data-efficient manner.
In some applications, edge learning is experiencing a shift in focusing from conventional learning from scratch to new two-stage learning unifying pre-training and task-specific fine-tuning. This paper considers the problem of joint communication and computation resource management in a two-stage edge learning system. In this system, model pre-training is first conducted at an edge server via centralized learning on local pre-stored general data, and then task-specific fine-tuning is performed at edge devices based on the pre-trained model via federated edge learning. For the two-stage learning model, we first analyze the convergence behavior (in terms of the average squared gradient norm bound), which characterizes the impacts of various system parameters such as the number of learning rounds and batch sizes in the two stages on the convergence rate. Based on our analytical results, we then propose a joint communication and computation resource management design to minimize an average squared gradient norm bound, subject to constraints on the transmit power, overall system energy consumption, and training delay. The decision variables include the number of learning rounds, batch sizes, clock frequencies, and transmit power control for both pre-training and fine-tuning stages. Finally, numerical results are provided to evaluate the effectiveness of our proposed design. It is shown that the proposed joint resource management over the pre-training and fine-tuning stages well balances the system performance trade-off among the training accuracy, delay, and energy consumption. The proposed design is also shown to effectively leverage the inherent trade-off between pre-training and fine-tuning, which arises from the differences in data distribution between pre-stored general data versus real-time task-specific data, thus efficiently optimizing overall system performance.
This paper investigates the energy-efficient hybrid beamforming design for a multi-functional integrated sensing, communications, and powering (ISCAP) system. In this system, a base station (BS) with a hybrid analog-digital (HAD) architecture sends unified wireless signals to communicate with multiple information receivers (IRs), sense multiple point targets, and wirelessly charge multiple energy receivers (ERs) at the same time. To facilitate the energy-efficient design, we present a novel HAD architecture for the BS transmitter, which allows dynamic on-off control of its radio frequency (RF) chains and analog phase shifters (PSs) through a switch network. We also consider a practical and comprehensive power consumption model for the BS, by taking into account the power-dependent non-linear power amplifier (PA) efficiency, and the on-off non-transmission power consumption model of RF chains and PSs. We jointly design the hybrid beamforming and dynamic on-off control at the BS, aiming to minimize its total power consumption, while guaranteeing the performance requirements on communication rates, sensing Cram\'er-Rao bound (CRB), and harvested power levels. The formulation also takes into consideration the per-antenna transmit power constraint and the constant modulus constraints for the analog beamformer at the BS. The resulting optimization problem for ISCAP is highly non-convex. Please refer to the paper for a complete abstract.
Future sixth-generation (6G) systems are expected to leverage extremely large-scale multiple-input multiple-output (XL-MIMO) technology, which significantly expands the range of the near-field region. The spherical wavefront characteristics in the near field introduce additional degrees of freedom (DoFs), namely distance and angle, into the channel model, which leads to unique challenges in channel estimation (CE). In this paper, we propose a new sensing-enhanced uplink CE scheme for near-field XL-MIMO, which notably reduces the required quantity of baseband samples and the dictionary size. In particular, we first propose a sensing method that can be accomplished in a single time slot. It employs power sensors embedded within the antenna elements to measure the received power pattern rather than baseband samples. A time inversion algorithm is then proposed to precisely estimate the locations of users and scatterers, which offers a substantially lower computational complexity. Based on the estimated locations from sensing, a novel dictionary is then proposed by considering the eigen-problem based on the near-field transmission model, which facilitates efficient near-field CE with less baseband sampling and a more lightweight dictionary. Moreover, we derive the general form of the eigenvectors associated with the near-field channel matrix, revealing their noteworthy connection to the discrete prolate spheroidal sequence (DPSS). Simulation results unveil that the proposed time inversion algorithm achieves accurate localization with power measurements only, and remarkably outperforms various widely-adopted algorithms in terms of computational complexity. Furthermore, the proposed eigen-dictionary considerably improves the accuracy in CE with a compact dictionary size and a drastic reduction in baseband samples by up to 77%.
Federated learning (FL) allows collaborative machine learning training without sharing private data. While most FL methods assume identical data domains across clients, real-world scenarios often involve heterogeneous data domains. Federated Prototype Learning (FedPL) addresses this issue, using mean feature vectors as prototypes to enhance model generalization. However, existing FedPL methods create the same number of prototypes for each client, leading to cross-domain performance gaps and disparities for clients with varied data distributions. To mitigate cross-domain feature representation variance, we introduce FedPLVM, which establishes variance-aware dual-level prototypes clustering and employs a novel $\alpha$-sparsity prototype loss. The dual-level prototypes clustering strategy creates local clustered prototypes based on private data features, then performs global prototypes clustering to reduce communication complexity and preserve local data privacy. The $\alpha$-sparsity prototype loss aligns samples from underrepresented domains, enhancing intra-class similarity and reducing inter-class similarity. Evaluations on Digit-5, Office-10, and DomainNet datasets demonstrate our method's superiority over existing approaches.
The fusion of complementary multimodal information is crucial in computational pathology for accurate diagnostics. However, existing multimodal learning approaches necessitate access to users' raw data, posing substantial privacy risks. While Federated Learning (FL) serves as a privacy-preserving alternative, it falls short in addressing the challenges posed by heterogeneous (yet possibly overlapped) modalities data across various hospitals. To bridge this gap, we propose a Federated Multi-Modal (FedMM) learning framework that federatedly trains multiple single-modal feature extractors to enhance subsequent classification performance instead of existing FL that aims to train a unified multimodal fusion model. Any participating hospital, even with small-scale datasets or limited devices, can leverage these federated trained extractors to perform local downstream tasks (e.g., classification) while ensuring data privacy. Through comprehensive evaluations of two publicly available datasets, we demonstrate that FedMM notably outperforms two baselines in accuracy and AUC metrics.
Linguistic ambiguity is ubiquitous in our daily lives. Previous works adopted interaction between robots and humans for language disambiguation. Nevertheless, when interactive robots are deployed in daily environments, there are significant challenges for natural human-robot interaction, stemming from complex and unpredictable visual inputs, open-ended interaction, and diverse user demands. In this paper, we present SInViG, which is a self-evolving interactive visual agent for human-robot interaction based on natural languages, aiming to resolve language ambiguity, if any, through multi-turn visual-language dialogues. It continuously and automatically learns from unlabeled images and large language models, without human intervention, to be more robust against visual and linguistic complexity. Benefiting from self-evolving, it sets new state-of-the-art on several interactive visual grounding benchmarks. Moreover, our human-robot interaction experiments show that the evolved models consistently acquire more and more preferences from human users. Besides, we also deployed our model on a Franka robot for interactive manipulation tasks. Results demonstrate that our model can follow diverse user instructions and interact naturally with humans in natural language, despite the complexity and disturbance of the environment.
Interactive visual grounding in Human-Robot Interaction (HRI) is challenging yet practical due to the inevitable ambiguity in natural languages. It requires robots to disambiguate the user input by active information gathering. Previous approaches often rely on predefined templates to ask disambiguation questions, resulting in performance reduction in realistic interactive scenarios. In this paper, we propose TiO, an end-to-end system for interactive visual grounding in human-robot interaction. Benefiting from a unified formulation of visual dialogue and grounding, our method can be trained on a joint of extensive public data, and show superior generality to diversified and challenging open-world scenarios. In the experiments, we validate TiO on GuessWhat?! and InViG benchmarks, setting new state-of-the-art performance by a clear margin. Moreover, we conduct HRI experiments on the carefully selected 150 challenging scenes as well as real-robot platforms. Results show that our method demonstrates superior generality to diversified visual and language inputs with a high success rate. Codes and demos are available at https://github.com/jxu124/TiO.
Videos captured from multiple viewpoints can help in perceiving the 3D structure of the world and benefit computer vision tasks such as action recognition, tracking, etc. In this paper, we present a method for self-supervised learning from synchronized multi-view videos. We use a cross-view reconstruction task to inject geometry information in the model. Our approach is based on the masked autoencoder (MAE) framework. In addition to the same-view decoder, we introduce a separate cross-view decoder which leverages cross-attention mechanism to reconstruct a target viewpoint video using a video from source viewpoint, to help representations robust to viewpoint changes. For videos, static regions can be reconstructed trivially which hinders learning meaningful representations. To tackle this, we introduce a motion-weighted reconstruction loss which improves temporal modeling. We report state-of-the-art results on the NTU-60, NTU-120 and ETRI datasets, as well as in the transfer learning setting on NUCLA, PKU-MMD-II and ROCOG-v2 datasets, demonstrating the robustness of our approach. Code will be made available.