Electroencephalography (EEG)-based applications in Brain-Computer Interfaces (BCIs), neurological disease diagnosis, rehabilitation, and other areas rely on the utilization of extensive data for model development. Nevertheless, this raises concerns regarding storage and privacy, since model development needs a significant amount of data, and EEG sharing discloses sensitive information such as identity and health. To address this challenging problem, we provide the paradigm of EEG condensation, aiming to generate a synthetic sample set that is highly information-concentrated yet not visually similar. Correspondingly, we propose a novel dataset condensation framework where the knowledge of the original EEG dataset is condensed into diffusion models, named EEGCiD. Specifically, EEGCiD first utilizes a deterministic denoising diffusion implicit model (DDIM) to store the information of the original dataset and optimizes the condensation latent codes z to obtain the EEG condensation dataset. Further, to enhance the modeling of EEG knowledge in DDIM, we design a transformer architecture incorporating the spatial and temporal self-attention block (STSA) to replace the traditional U-Net backbone. In the condensation phase, EEGCiD randomly initializes a subset of samples from the original dataset to obtain the condensation latent codes z through the forward process in DDIM. Then, it optimizes z by matching the feature distributions in multiple EEG decoding models between the synthetic samples and the original dataset. Extensive experiments across three EEG datasets demonstrate that the condensation dataset from the proposed model not only achieves superior classification performance with limited sample sizes, but also effectively prevents membership inference attacks (MIA). Note to Practitioners—This paper aims to investigate a novel EEG generation paradigm that extracts representative synthetic samples from large-scale datasets. Existing studies in EEG generation primarily concentrate on generating real-like signals, and some work claims that the generated EEG can serve as a substitute for the original dataset to achieve privacy preservation. In the EEGCiD framework, the deterministic DDIM is pre-trained with the original dataset to store the knowledge. Besides, an ensemble feature matching strategy is proposed to condense the information from the original dataset into a small latent code set. Experiments on three datasets demonstrate that EEGCiD addresses two fundamental challenges: 1) obtaining superior classification performance within a small dataset (limited storage capacity); 2) avoiding potential privacy issues during EEG sharing and transmission.

Optimizing the milling parameters for complex surface can improve the machining quality. However, existing methods are oriented toward a simple machining path and single machining process. In this paper, a milling parameter optimization method for complex surface is proposed by combining the data-driven models and physical constraints. Initially, the chatter indicator is derived from the tool vibration using Variational Mode Decomposition. Subsequently, multiple few-shot prediction models are developed based on real machining data using deep neural networks. A mathematical optimization model is then constructed by combining multiple prediction models and physical constraints, which aims to minimize the machining failure rate and maximize the material removal rate under multiple constraints. The spindle speed, feed speed, cutting depth, and path spacing are the optimization parameters. Finally, a Hypervolume-based Multi-Objective Optimization (HMOO) algorithm is proposed to solve the optimization model. The solution set produced by HMOO exhibits superior convergence and diversity compared to the S-Metric Selection Based Evolutionary Multi-Objective Algorithm (SMS-MOEA). In experiments, a five-axis machine tool is employed to mill turbine blades with complex surface. Experimental results demonstrate that the proposed prediction models achieve higher accuracy than widely used regression algorithms. Integrating the high-precision prediction models with HMOO significantly enhances blade machining quality while guaranteeing reliable machining efficiency, resulting in a 5.25% reduction in the machining failure rate.Note to Practitioners—Methods designed for simple machining paths and single machining processes are inadequate for optimizing milling parameters in complex surfaces. Given the challenging machining characteristics and stringent precision demands of complex surfaces, this paper proposes a milling parameter optimization approach for complex surface by integrating data-driven models with real physical constraints. The approach employs neural network-based prediction models to map the relationships among milling parameters, machining conditions, and machining accuracy, while accounting for actual physical constraints in the optimization process. Our approach enhances the machining accuracy of complex surfaces while ensuring reliable machining efficiency.

Value function approximation, such as Q-learning, is widely used in the discrete control rather than the continuous one because the optimal action in the discrete setting is more easily selected. Optimizing the action is a non-convex optimization problem with respect to the complex value function. Some notable studies simplify the non-convex optimization problem by assuming the value function as quadratic in the actions or by discretizing the action space. However, the performance of the output policy will decline if these studies’ premises do not hold. In order to address the problem, we propose a framework that combines swarm intelligence algorithms with value-based Reinforcement Learning, where the swarm intelligence algorithms are employed to search for the optimal action with respect to the state and the value function. To ensure the correctness of this framework, we conditionally claim the convergence rate of swarm intelligence algorithms with high probability. We then implement it by searching the batch optimal actions to various states on the GPU platform for the batch training. Furthermore, we employ the population-based atomic actions for the compatibility with the existing related work about solving discrete control problems. Four classical control models and four robot simulation environments are utilized in the comparisons. According to empirical results, our framework outputs a policy comparable with that of the policy-based algorithms by 10% timesteps in the continuous control. Note to Practitioners—This paper is motivated by the exploration-exploitation dilemma of Reinforcement Learning to solve continuous control tasks. To balance the exploration and exploitation, the stochastic exploration and the prioritized exploration are roughly two feasible ways, where the prioritized one is a better choice due to the higher data efficiency than the stochastic one, e.g.$\varepsilon $-greedy. Normally, the prioritized exploration works well in the value-based Reinforcement Learning algorithms rather than the policy-based ones; meanwhile, the policy-based algorithms are more suitable to continuous control tasks than the value-based ones. To tackle this conflict, we especially design a particle swarm optimization to maximize the Q-value of action in Q-learning. Our design can be hybridized by various swarm intelligence and value-based Reinforcement Learning algorithms. Also, it can be embedded in most intelligent control systems easily. The aim of this study is to solve the continuous control tasks by value-based algorithms as the first step of applying the prioritized exploration. The simulative results verify the effectiveness and efficiency of our design.

Mobile edge computing (MEC) is an emerging computing paradigm that decreases the computing time and extends the lifespan of user equipments (UEs). In MEC, the computational tasks are offloaded from UEs to the base station (BS) at the edge of the network for processing. However, MEC cannot cope with environments where there are no BS or where communication facilities have been destroyed. In this paper, we study the problem of minimizing the energy consumption of UAV equipped with MEC servers as a mobile base station to serve users. The problem involves user offloading decision, UAV location and allocation with computational resources, and is a hybrid optimization problem with continuous and discrete variables. To address this problem, we propose a hybrid nature-inspired optimization algorithm (HNIO) and its version for discrete optimization, where HNIO incorporates mutation and population diversity detection mechanisms to boost its global optimization capability, and we design a probabilistic selection-based coding strategy for the discrete optimization version. The experimental study is conducted based on ten cases with different numbers of UEs. Comparing HNIO with several other state-of-the-art optimization algorithms, it is concluded from the Friedman and Wilcoxon’s test of the experimental results that HNIO shows better precision and stability in nine out of the ten cases with higher number of UEs.

The multimission selective maintenance problem (MSMP) for repairable systems has received increasing attention in recent years. The problem amounts to selecting a subset of feasible maintenance actions, in view of the resource limitations. For considering the realistic case of the imperfect maintenance, this article introduces a hybrid imperfect maintenance model, which is more realistic to evaluate the system reliability. The challenge of solving such kind of problems lies not only in the reliability estimation, but also in the solution method of the maintenance selection. Such decision-making problem can be effectively formulated using the Markov decision process, but it is difficult to apply current methods for solving the engineering systems with large action decision spaces. In order to solve this issue, this work puts forth a novel hybrid algorithm for the MSMP in a large multicomponent system. In the proposed method, a discrete differential evolution algorithm is developed for searching the optimal maintenance action in large-scale discrete action spaces and the deep Q-network method is utilized to approximate the effectiveness of maintenance actions and facilitate the agent training. The experiments, based on a large-scale coal transportation system, verify the effectiveness of the proposed method compared with LSDQN and differential evolution.

This letter focuses on the problem of lifelong classification in the open world, the goal of which is to achieve an endless process of learning. However, incremental data sets (like the streaming data) in the open world, where the new classes may be emerging, are unsuited for classical classification methods. For addressing this problem, existing methods usually retrain the whole observed data sets with the complex computation and the expensive storage cost. This letter attempts to improve the performance of classification in the open world and decomposes the problem into three subproblems: (1) to reject unknown instances, (2) to classify accepted instances, and (3) to cut the cost of learning. Rejecting unknown instances refers to recognize those instances whose classes are unknown according to the learner, which could reduce the computation of the retraining process and eliminate the storage of historical data sets. We employ outlier detection for rejecting instances and a variant artificial neural network for classifying with fewer weights. Results on several experiments show that the work is effective. Source code can be found at https://github.com/wangbi1988/Lifelong-learning-in-Open-World-Classification.