Gamma belief network (GBN), widely viewed as deep probabilistic topic models, has demonstrated its potential for uncovering multi-layer interpretable latent representations from text corpora. Its notable performance in document modeling largely arises from the expressive nature of gamma-distributed latent variables, which naturally capture sparsity, nonnegativity, skewness, heavy-tailed pattens, and from their seamless extension to multi-layer hierarchical structures. However, existing GBN and its variations are constrained by linear generative model, thereby limiting their expressiveness and applicability. To address this limitation, we introduce Generalized Gamma Belief Network (Generalized GBN), which extends original linear generative model to a more expressive non-linear generative model. Since parameters of Generalized GBN no longer possess an analytic conditional posterior, we further propose an upward-downward Weibull inference network to approximate posterior distribution of latent variables. The parameters of both generative model and inference network are jointly trained within variational inference framework. In addition, we provide theoretical analyses that demonstrate the effectiveness of Generalized GBN in modeling data variability and achieving disentangled representations. The former benefit arises from its hierarchical latent-variable structure, while the latter stems from its inherent ability to model sparsity. Finally, we conduct comprehensive experiments on both expressivity and disentangled representation learning tasks to evaluate the performance of Generalized GBN against Gaussian variational autoencoders serving as strong baseline models.

The influence function serves as an efficient post-hoc interpretability tool that quantifies the impact of training data modifications on model parameters, enabling enhanced model performance, improved generalization, and interpretability insights without the need for expensive retraining processes. Recently, Multivariate Time Series (MTS) analysis has become an important yet challenging task, attracting significant attention. While channel extremely matters to MTS tasks, channel-centric methods are still largely under-explored for MTS. Particularly, no previous work studied the effects of channel information of MTS in order to explore counterfactual effects between these channels and model performance. To fill this gap, we propose a novel Channel-wise Influence (ChInf) method that is the first to estimate the influence of different channels in MTS. Based on ChInf, we naturally derived two channel-wise algorithms by incorporating ChInf into classic MTS tasks. Extensive experiments demonstrate the effectiveness of ChInf and ChInf-based methods in critical MTS analysis tasks, such as MTS anomaly detection and MTS data pruning. Specifically, our ChInf-based methods rank top-1 among all methods for comparison, while previous influence functions do not perform well on MTS anomaly detection tasks and MTS data pruning problem. This fully supports the superiority and necessity of ChInf.

Orbital angular momentum (OAM) in electromagnetic (EM) waves can significantly enhance spectrum efficiency in wireless communications without requiring additional power, time, or frequency resources. Different OAM modes in EM waves create orthogonal channels, thereby improving spectrum efficiency. Additionally, OAM waves can more easily maintain orthogonality in line-of-sight (LOS) transmissions, offering an advantage over multiple-input and multiple-output (MIMO) technology in LOS scenarios. However, challenges such as divergence and crosstalk hinder OAM’s efficiency. Additionally, channel modeling for OAM transmissions is still limited. A reliable channel model with balanced accuracy and complexity is essential for further system analysis. In this paper, we present a quasi-deterministic channel model for OAM channels in the 5.8 GHz and 28 GHz bands based on measurement data. Accurate measurement, especially at high frequencies like millimeter bands, requires synchronized RF channels to maintain phase coherence and purity, which is a major challenge for OAM channel measurement. To address this, we developed an 8-channel OAM generation device at 28 GHz to ensure beam integrity. By measuring and modeling OAM channels at 5.8 GHz and 28 GHz with a modified 3D geometric-based stochastic model (GBSM), this study provides insights into OAM channel characteristics, aiding simulation-based analysis and system optimization.

The forecasting of Multivariate Time Series (MTS) has long been an important but challenging task. Due to the non-stationary problem across long-distance time steps, previous studies primarily adopt stationarization method to attenuate the non-stationary problem of original series for better predictability. However, existed methods always adopt the stationarized series, which ignore the inherent non-stationarity, and have difficulty in modeling MTS with complex distributions due to the lack of stochasticity. To tackle these problems, we first develop a powerful hierarchical probabilistic generative module to consider the non-stationarity and stochastity characteristics within MTS, and then combine it with transformer for a well-defined variational generative dynamic model named Hierarchical Time series Variational Transformer (HTV-Trans), which recovers the intrinsic non-stationary information into temporal dependencies. Being an powerful probabilistic model, HTV-Trans is utilized to learn expressive representations of MTS and applied to the forecasting tasks. Extensive experiments on diverse datasets show the efficiency of HTV-Trans on MTS forecasting tasks.

Flexible electronics can conform to curved surfaces, playing an essential role in biomedical and industrial electronics. Real-time and wireless data acquisition is crucial for flexible electronics to achieve intelligent functions. Previous works mainly focus on improving the flexibility of front-end sensors, whereas back-end integrated circuits for data acquisition become the bottleneck for developing flexible electronic systems. This work reports flexible integrated circuits (FIC) with stress-minimized layout and ultra-thin chips. 25 μm-thick chips are integrated into the FIC, significantly improving the flexibility of the FIC system. A dispersed layout of chips is proposed to reduce additional stress on the substrate caused by chips during deformation. An optimization model based on mechanical principles is also proposed to minimize the stress endured by ultra-thin chips. The FIC adopts a system-level package method to promote its mechanical reliability essentially. FIC is small, lightweight, and can monitor multiple signals in real-time. The demonstration of FIC opens up new prospects for flexible integrated electronics in biomedical and industrial fields.

Orbital angular momentum (OAM) carried electromagnetic waves have the potential to improve spectrum efficiency in optical and radio-frequency communications due to the orthogonal wavefronts of different OAM modes. However, OAM beams are vortically hollow and divergent, which significantly decreases the capacity of OAM transmissions. In addition, unaligned transceivers in OAM transmissions can result in a high bit error rate (BER). The Talbot effect is a self-imaging phenomenon that can be used to generate optical or radio-frequency OAM beams with periodic repeating structures at multiples of a certain distance along the propagation direction. These periodic structures make it unnecessary for the transceiver antennas to be perfectly aligned and can also alleviate the hollow divergence of OAM beams. In this paper, we propose Talbot-effect-based fractal OAM generation and detection schemes using a uniform circular array (UCA) to significantly improve capacity and BER performance in unaligned OAM transmissions. We first provide a brief overview of fractal OAM. Then, we propose the fractal OAM beam generation and detection schemes. Numerical analysis and simulations verify the effectiveness of our proposed fractal OAM generation scheme and also demonstrate improved capacity and BER performance compared to normal OAM transmissions. We also analyze how the receive UCA radius and the distance between the UCAs impact the capacity and BER performances.

In heavy-duty series hybrid electric vehicles (SHEVs), engine-generator set (EGS) functions as the main power source for propulsion. However, limitation of engine power per liter and delayed computation of control algorithm result in the hysteretic response of EGS to high demand power. It leads to deteriorating operation of powertrain. Thus, challenging technical issue lies in achieving stable powertrain operation which is difficult to describe precisely by real-time control. In this work, an efficient power control scheme for heavy-duty HEV with online optimized variable universe fuzzy system is proposed. First, a splitting sequential clustering quadratic programming (SSCQP) algorithm is designed to solve power distribution and achieve real-time control. The original subproblem is split into two subproblems with smaller scale to obtain iterative points. And clustering algorithm is introduced to gather up the points to improve the termination criterion. It turns to skip unnecessary short step in the iteration which fails to obtain sufficient descent. Then, the online optimized variable universe fuzzy system is established to achieve rapid response of EGS by adjusting power distribution. In this system, online optimization of membership function distribution parameters is considered. The optimization is constructed on real-time membership overlap degree and central value of fuzzy system rather than the traditional off-line optimization using posterior information of vehicle. Finally, effectiveness of proposed scheme is validated both in simulation test and hardware-in-loop test. The results reveal that stable power output is maintained and calculation time is decreased by 40.9%, 46.0% under two driving cycles.

This study proposes a sequential clustering quadratic programming (SCQP) method for the energy management strategies of plug-in hybrid electric vehicles (PHEVs). In this method, the clustering algorithm is introduced to gather up the points with a smaller iteration step size in the iteration process. The clustering results are utilized to design the termination criterion based on the distance between the cluster centers of various iteration domains. In the case that the distance varies within the preset range, it indicates that the current iteration point is sufficiently close to the optimal point. So that the criterion turns to terminate the computation to reduce unnecessary iteration steps. To analyze the convergence of the method with the designed criterion, the mathematical illustrations are proposed. In the mathematical illustrations, the monotonicity of the clustering objective function is firstly given. Then, the theorem of feasibility for the solution obtained by the designed criterion is proved. On the basis of aforementioned conclusions, the convergence of the SCQP method is obtained. Finally, the performance of the proposed method is validated both in simulation test and hardware-in-loop (HIL) test. The simulation results reveal that the PHEV achieves 8.81% and 7.74% less fuel consumption under two driving cycles. And the average iteration number of the proposed method is obviously reduced compared with the conventional SQP. The HIL results reveal that the proposed strategy exhibits similar performance in both real controller and simulation. The energy saving and real-time performance can be verified.

Orbital angular momentum (OAM) carried electro-magnetic waves can be used for optical and radio-frequency communications to improve spectrum efficiency thanks to the orthogonal wavefronts of different OAM modes. However, OAM beams are vortically hollow and divergent, which seriously decreases the capacity for OAM transmissions. Moreover, unaligned-transceiver-based OAM transmissions can lead to a high bit error rate. In this paper, we propose the fractal OAM generation and detection schemes, which can alleviate the hollow divergence of OAM beams and greatly improve the capacity performance for unaligned OAM transmissions. We first briefly introduce the fractal OAM phenomenon. Then, we propose the fractal OAM beam generation and detection schemes. Simulations verify our proposed fractal OAM generation scheme. The improved capacity performance of our proposed fractal OAM compared with normal OAM transmissions are also validated via simulations.