Causal discovery from observational data is a fundamental task in artificial intelligence, with far-reaching implications for decision-making, predictions, and interventions. Despite significant advances, existing methods can be broadly categorized as constraint-based or score-based approaches. Constraint-based methods offer rigorous causal discovery but are often hindered by small sample sizes, while score-based methods provide flexible optimization but typically forgo explicit conditional independence testing. This work explores a third avenue: developing differentiable $d$-separation scores, obtained through a percolation theory using soft logic. This enables the implementation of a new type of causal discovery method: gradient-based optimization of conditional independence constraints. Empirical evaluations demonstrate the robust performance of our approach in low-sample regimes, surpassing traditional constraint-based and score-based baselines on a real-world dataset. Code implementing the proposed method is publicly available at [https://github.com/PurdueMINDS/DAGPA](https://github.com/PurdueMINDS/DAGPA).

Pantana phyllostachysaeChao (P. phyllostachysae) is a destructive leaf-eating pest that poses a significant threat to the health of bamboo forests and the bamboo industry. However, the spatial and temporal spread mechanisms of this pest are still unclear. To better understand and predict the spread of this pest, we used Sentinel-2A/B images from the pest detection period of 2018 to 2021, to identify association factors from five dimensions, including forest stand, meteorology, topography, pest sources, and human environment factors. The association factor sets for the spread ofP. phyllostachysaewere established under both existence and non-existence pest control scenarios. The extreme gradient boosting (XGBoost) model was employed to derive conversion rules for the respective spread models, enabling the determination of suitability probabilities for both healthy and damaged bamboo forests. These probabilities were then utilized in conjunction with cellular automata (CA) to simulate the spread ofP. phyllostachysaeunder two scenarios. The results showed that the OA and Kappa reached more than 85% and 0.7 in both scenarios, respectively. Meanwhile, the division of pest control scenarios and the selection of XGBoost both help to improve the spreading simulation accuracy. Our models effectively coupled the research results of leaf hosts of different damage levels, simulated the spread ofP. phyllostachysae, and identified the dynamic mechanisms of the pest’s spread. These findings provide decision support for interrupting the spread path of the pest and achieving precise control, thus safeguarding forest ecological security.

There are many impurities in the microscopic images of extracted microsporidia samples of Bombyx mori pebrine, and Bombyx mori pebrine with elliptical symmetric shape has certain fluidity and obvious stratification. Traditional focusing methods cannot accurately locate the main regions of microsporidia images, and the focusing effect is poor. On this basis, an automatic focusing method combining the microsporidia image features and the evaluation and determination of maximum gradient direction is proposed. First, the HSV color space with stable color information is used to extract the suspected positions of microsporidia targets, so that the interference of some impurities under complex backgrounds is removed and the redundancy of image content calculation is reduced. Then, combined with the light green features of Bombyx mori pebrine, the G-component gray image of microsporidia in the RGB color space is used to extract the significant gradient region. A dynamic focus window is constructed to accurately locate the target region and reduce the influence of microsporidia flow on the focus evaluation function and the bimodal interference caused by impurities. Finally, the maximum second-order difference is obtained through the four-dimensional gradient distribution, and the focus sharpness evaluation function is formulated to adapt to the microsporidia shape and improve the sensitivity of the focus function. The experiments show that under the dynamic window of microsporidia color gradient of different samples, the sharpness ratio and the highest sensitivity factor of the focus evaluation function proposed in this paper can reach 0.06341 and 0.95, respectively. It can meet the accurate and sensitive autofocus of microscopic images of color microsporidia samples under complex backgrounds.

With the rapid development of artificial intelligence (AI) in recent years, fault diagnostics for industrial applications have leaped toward partially or fully automatic provided by the capability of analyzing massive condition monitoring data from sensors and actuators. Generally, AI-based fault diagnostics can achieve high accuracy when failure types appear in training dataset and testing dataset are the same. These diagnostic methods could be invalidated for applications dealing with unprecedented faults because the pretrained classifier for diagnostics tends to misclassify the novel instances into existing known classes. In order to address these limitations of conventional diagnostic approaches, we propose a unified diagnostics framework that can achieve novel fault detection and known fault classification tasks together. Through jointly training a variational autoencoder and a deep neural networks classifier, we convert the original entangled raw data into latent variables with Gaussian probabilistic distributions in the latent space and utilize the probabilistic latent variables to detect novel samples against known fault classes or classify them into one of the existing fault classes if they are not novel. The effectiveness of our proposed joint-training framework is validated through experimental studies on two different bearing datasets. Compared with the state-of-the-art methods in the literature, our unified framework is able to not only accurately detect the novel fault classes but also achieve high classification accuracy of known fault classes.

Augmented reality (AR) is on the rise with consumer-level head-mounted displays (HMDs) becoming available in recent years. Text entry is an essential activity for AR systems, but it is still relatively underexplored. Although it is possible to use a physical keyboard to enter text in AR systems, it is not the most optimal and ideal way because it confines the uses to a stationary position and within indoor environments. Instead, a virtual keyboard seems more suitable. Text entry via virtual keyboards requires a pointing method and a selection mechanism. Although there exist various combinations of pointing+selection mechanisms, it is not well understood how well suited each combination is to support fast text entry speed with low error rates and positive usability (regarding workload, user experience, motion sickness, and immersion). In this research, we perform an empirical study to investigate user preference and text entry performance of four pointing methods (Controller, Head, Hand, and Hybrid) in combination with two input mechanisms (Swype and Tap). Our research represents a first systematic investigation of these eight possible combinations. Our results show that Controller outperforms all the other device-free methods in both text entry performance and user experience. However, device-free pointing methods can be usable depending on task requirements and users' preferences and physical condition.

Energy harvesting is becoming an attractive option of energy supply for wireless networks as it can effectively reduce capital expenditure (CAPEX) and operational expenditure (OPEX). In this paper, an energy efficient radio resource optimization algorithm is proposed for a two-tier heterogeneous cloud radio access network (H-CRAN) where macro cells are empowered by conventional grid power and remote radio heads (RRH) are empowered by renewable energy sources. The resource allocation optimization is firstly formulated as a mixed integer programming problem, which is NP-hard. Therefore, an equivalent green power utilization maximization problem is formulated, and solved by Lagrange dual decomposition method. Numerical results show that the proposed algorithm can increase the utilization of the green power harvested from the renewable energy sources. This, in turn, leads to reduced grid power consumption compared to the baseline algorithms.

One of key 5G scenarios is that device-to-device (D2D) and massive multiple-input multiple-output (MIMO) will be co-existed. However, interference in the uplink D2D underlaid massive MIMO cellular networks needs to be coordinated, due to the vast cellular and D2D transmissions. To this end, this paper introduces a spatially dynamic power control solution for mitigating the cellular-to-D2D and D2D-to-cellular interference. In particular, the proposed D2D power control policy is rather flexible including the special cases of no D2D links or using maximum transmit power. Under the considered power control, an analytical approach is developed to evaluate the spectral efficiency (SE) and energy efficiency (EE) in such networks. Thus, the exact expressions of SE and EE for a cellular user or D2D transmitter are derived, which quantify the impacts of key system parameters such as massive MIMO antennas and D2D density. Numerical results corroborate our analysis and show that the proposed power control solution can efficiently mitigate interference between the cellular and D2D tier.

One of the key 5G scenarios is that device-to-device (D2D) and massive multiple-input multiple-output (MIMO) will be co-existed. However, interference in the uplink D2D underlaid massive MIMO cellular networks needs to be coordinated, due to the vast cellular and D2D transmissions. To this end, this paper introduces a spatially dynamic power control solution for mitigating the cellular-to-D2D and D2D-to-cellular interference. In particular, the proposed D2D power control policy is rather flexible, including the special cases of no D2D links or using maximum transmit power. Under the considered power control, an analytical approach is developed to evaluate the spectral efficiency (SE) and energy efficiency (EE) in such networks. Thus, the exact expressions of SE for a cellular user or D2D transmitter are derived, which quantify the impacts of key system parameters, such as massive MIMO antennas and D2D density. Moreover, the D2D scale properties are obtained, which provide the sufficient conditions for achieving the anticipated SE. Numerical results corroborate our analysis and show that the proposed power control solution can efficiently mitigate interference between the cellular and the D2D tier. The results demonstrate that there exists the optimal D2D density for maximizing the area SE of D2D tier. In addition, the achievable EE of a cellular user can be comparable with that of a D2D user.

This paper considers the massive multiple-input multiple-output (MIMO) enabled heterogeneous cloud radio access network (C-RAN), in which both remote radio heads (RRHs) and massive MIMO macrocell base stations (BS) are deployed to potentially accomplish high throughput and energy efficiency (EE). In this network, the soft fractional frequency reuse (S-FFR) is employed to mitigate the inter-tier interference. We develop a tractable analytical approach to evaluate the throughput and EE of the entire network, which can well predict the impacts of the key system parameters such as number of macrocell BS antennas, RRH density, and S-FFR factor, etc. Our results demonstrate that massive MIMO is still a powerful tool for improving the throughput of the heterogeneous C-RAN while RRHs are capable of achieving higher EE. The impact of S-FFR on the network throughput is dependent on the density of RRHs. Furthermore, more radio resources allocated to the RRHs can greatly improve the EE of the network.

This paper exploits the potential of massive multiple input multiple output (MIMO) in K-tier heterogeneous cellular networks (HCNs), to enhance the data rate for 5G. In such a network, macro base stations (MBSs) are equipped with large number of antennas and support multi-user transmission. We first examine the impact of massive MIMO on user association in K-tier HCNs. Exact and asymptotic expressions for the probability of a user being associated with a macro cell or a small cell are derived. Based on the asymptotic analysis, the impacts of system parameters such as tier's density and BS transmit power on user association are explicitly identified. Furthermore, we derive the coverage probability and rate of the proposed network. Numerical results corroborate our analysis and show that the implementation of massive MIMO in macro cells can significantly enhance the performance of HCNs in terms of coverage and rate. A guideline for practical cellular deployment is reached that MBSs with large antenna arrays can decrease the demands for small cells.

Base station (BS) sleeping has been proved to be an effective technique for saving energy consumption in cellular networks. However, BSs in sleeping mode might cause coverage holes, which have a negative impact on the connectivity of the network. In order to overcome this problem, we propose a combined coordinated multi-point (CoMP) transmission and BS sleeping scheme under the heterogeneous networks (HetNets) scenario. The proposed scheme aims at improving energy efficiency as well as increasing coverage probability. In this paper, stochastic geometry analysis is adopted for evaluating the performance of the proposed scheme, instead of the conventional hexagonal grid based approach. Impact of CoMP on energy efficiency in HetNets with a random on/off strategy applied at macro base stations (MBSs) is examined thoroughly. We derived two performance indicators which are coverage probability and energy efficiency in a two-tier HetNets scenario. Numerical results confirm that the combined CoMP and BS sleeping can improve the energy efficiency as well as increase the coverage probability compared with implementing BS sleeping only. The impact of the density ratio of MBSs to Pico BSs (PBSs) on energy efficiency and coverage probability is also quantified.