Differentially Private Bayesian Neural Networks on Accuracy, Privacy and Reliability

Bayesian neural network (BNN) allows for uncertainty quantification in prediction, offering an advantage over regular neural networks that has not been explored in the differential privacy (DP) framework. We fill this important gap by leveraging recent development in Bayesian deep learning and privacy accounting to offer a more precise analysis of the trade-off between privacy and accuracy in BNN. We propose three DP-BNNs that characterize the weight uncertainty for the same network architecture in distinct ways, namely DP-SGLD (via the noisy gradient method), DP-BBP (via changing the parameters of interest) and DP-MC Dropout (via the model architecture). Interestingly, we show a new equivalence between DP-SGD and DP-SGLD, implying that some non-Bayesian DP training naturally allows for uncertainty quantification. However, the hyperparameters such as learning rate and batch size, can have different or even opposite effects in DP-SGD and DP-SGLD. Extensive experiments are conducted to compare DP-BNNs, in terms of privacy guarantee, prediction accuracy, uncertainty quantification, calibration, computation speed, and generalizability to network architecture. As a result, we observe a new tradeoff between the privacy and the reliability. When compared to non-DP and non-Bayesian approaches, DP-SGLD is remarkably accurate under strong privacy guarantee, demonstrating the great potential of DP-BNN in real-world tasks.

Effect of Input-output Randomness on Gameplay Satisfaction in Collectable Card Games

Randomness is an important factor in games, so much so that some games rely almost purely on it for its outcomes and increase players engagement with them. However, randomness can affect the game experience depending on when it occurs in a game, altering the chances of planning for a player. In this paper, we refer to it as input-output randomness. Input-output randomness is a cornerstone of collectable card games like Hearthstone, in which cards are drawn randomly (input randomness) and have random effects when played (output randomness). While the topic might have been commonly discussed by game designers and be present in many games, few empirical studies have been performed to evaluate the effects of these different kinds of randomness on the players satisfaction. This research investigates the effects of input-output randomness on collectable card games across four input-output randomness conditions. We have developed our own collectable card game and experimented with the different kinds of randomness with the game. Our results suggest that input randomness can significantly impact game satisfaction negatively. Overall, our results present helpful considerations on how and when to apply randomness in game design when aiming for players satisfaction.

Hypernetwork Dismantling via Deep Reinforcement Learning

Network dismantling aims to degrade the connectivity of a network by removing an optimal set of nodes and has been widely adopted in many real-world applications such as epidemic control and rumor containment. However, conventional methods usually focus on simple network modeling with only pairwise interactions, while group-wise interactions modeled by hypernetwork are ubiquitous and critical. In this work, we formulate the hypernetwork dismantling problem as a node sequence decision problem and propose a deep reinforcement learning (DRL)-based hypernetwork dismantling framework. Besides, we design a novel inductive hypernetwork embedding method to ensure the transferability to various real-world hypernetworks. Generally, our framework builds an agent. It first generates small-scale synthetic hypernetworks and embeds the nodes and hypernetworks into a low dimensional vector space to represent the action and state space in DRL, respectively. Then trial-and-error dismantling tasks are conducted by the agent on these synthetic hypernetworks, and the dismantling strategy is continuously optimized. Finally, the well-optimized strategy is applied to real-world hypernetwork dismantling tasks. Experimental results on five real-world hypernetworks demonstrate the effectiveness of our proposed framework.

PersonalityGate: A General Plug-and-Play GNN Gate to Enhance Cascade Prediction with Personality Recognition Task

Cascade prediction estimates the size or the state of a cascade from either microscope or macroscope. It is of paramount importance for understanding the information diffusion process such as the spread of rumors and the propagation of new technologies in social networks. Recently, instead of extracting hand-crafted features or embedding cascade sequences into feature vectors for cascade prediction, graph neural networks (GNNs) are introduced to utilize the network structure which governs the cascade effect. However, these models do not take into account social factors such as personality traits which drive humans participation in the information diffusion process. In this work, we propose a novel multitask framework for enhancing cascade prediction with a personality recognition task. Specially, we design a general plug-and-play GNN gate, named PersonalityGate, to couple into existing GNN-based cascade prediction models to enhance their effectiveness and extract individuals personality traits jointly. Experimental results on two real-world datasets demonstrate the effectiveness of our proposed framework in enhancing GNN-based cascade prediction models and in predicting individuals personality traits as well.

Heterogeneous Information Network-based Interest Composition with Graph Neural Network for Recommendation

Heterogeneous information network (HIN) is widely applied to recommendation systems due to its capability of modeling various auxiliary information with meta-path. However, existing HIN-based recommendation models usually fuse the information from various meta-paths by simple weighted sum or concatenation, which limits the improvement of performance because it lacks the capability of interest compositions among meta-paths. In this article, we propose a HIN-based Interest Composition model for Recommendation (HicRec). Specially, the representations of users and items are learnt with graph neural network on both graph structure and features in each meta-path, and a parameter sharing mechanism is utilized here to ensure the representations of users and items are in the same latent space. Then, users interest on each item from each pair of related meta-paths is calculated by a combination of the representations of users and items. The composed interests of users are obtained by a composition of them from both intra- and inter-meta-paths for recommendation. Extensive experiments are conducted on three real-world datasets and the results demonstrate the outperformance of our proposed HicRec against the baselines.

Causal Learner: A Toolbox for Causal Structure and Markov Blanket Learning

Causal Learner is a toolbox for learning causal structure and Markov blanket (MB) from data. It integrates functions for generating simulated Bayesian network data, a set of state-of-the-art global causal structure learning algorithms, a set of state-of-the-art local causal structure learning algorithms, a set of state-of-the-art MB learning algorithms, and functions for evaluating algorithms. The data generation part of Causal Learner is written in R, and the rest of Causal Learner is written in MATLAB. Causal Learner aims to provide researchers and practitioners with an open-source platform for causal learning from data and for the development and evaluation of new causal learning algorithms. The Causal Learner project is available at http://bigdata.ahu.edu.cn/causal-learner.

HyperCI: A higher order collective influence measure for hypernetwork dismantling

Network dismantling aims to scratch the network into unconnected fragments by removing an optimal set of nodes and has been widely adopted in many real-world applications such as epidemic control and rumor containment. However, conventional methods often disassemble the system from the perspective of classic networks, which have only pairwise interactions, and often ignored the more ubiquitous and nature group-wise interactions modeled by hypernetwork. Moreover, a simple network cant describe the collective behavior of multiple objects, it is necessary to solve related problems through hypernetwork dismantling. In this work, we designed a higher order collective influence measure to identify key node sets in hypernetwork. It comprehensively consider the environment in which the target node is located and its own characteristics to determine the importance of the node, so as to dismantle the hypernetwork by removing these selected nodes. Finally, we used the method to carry out a series of real-world hypernetwork dismantling tasks. Experimental results on five real-world hypernetworks demonstrate the effectiveness of our proposed measure.

Observation of optical vortices in momentum space

Vortex, the winding of a vector field in two dimensions, has its core the field singularity and its topological charge defined by the quantized winding angle of the vector field. Vortices are one of the most fundamental topological excitations in nature, widely known in hair whorls as the winding of hair strings, in fluid dynamics as the winding of velocities, in angular-momentum beams as the winding of phase angle and in superconductors and superfluids as the winding of order parameters. Nevertheless, vortices have hardly been observed other than those in the real space. Although band degeneracies, such as Dirac cones, can be viewed as momentum-space vortices in their mathematical structures, there lacks a well-defined physical observable whose winding number is an arbitrary signed integer. Here, we experimentally observed momentum-space vortices as the winding of far-field polarization vectors in the Brillouin zone (BZ) of periodic plasmonic structures. Using a home-made polarization-resolved momentum-space imaging spectroscopy, we completely map out the dispersion, lifetime and polarization of all radiative states at the visible wavelengths. The momentum space vortices were experimentally identified by their winding patterns in the polarization-resolved iso-frequency contours and their diverging radiative quality factors. Such polarization vortices can exist robustly on any periodic systems of vectorial fields, while they are not captured by the existing topological band theory developed for scaler fields. This work opens up a promising avenue for exploring topological photonics in the momentum space, studying bound states in continuum (BICs), as well as for rendering and steering vector beams and designing high-Q plasmonic resonances.

A semi-empirical response function for Gamma-ray of Scintillation detector based on physical interaction mechanism

Scintillation detector has lower energy resolution for Gamma-ray as compared to semiconductor detector, better spectra analysis method is essential to traditional method. A model for describing the response function of scintillation detector over the range of incident Gamma-ray energies between 0.5 and 1.5 MeV has been established and applied to fitting radiation sources spectra. Each function form for describing the feature of Gamma-ray spectra are based on the analysis of fundamental interaction mechanism. These functions are combined to form a DRF model to fit experiment spectra by weighted least squares fitting method, parameters in this model are obtained simultaneously. Gaussian standard deviation can be calculated out by an individual procedure. Validity of the DRF model is demonstrated by fitting Co-60 and Cs-137 spectra measured by CsI(Tl) detector and comparing them to the normalized equivalent measured spectrum.