Jiahui Wei

Jiahui Wei, Marios Kountouris

We extend the Rate-Distortion-Perception (RDP) framework to the Rényi information-theoretic regime, utilizing Sibson's $α$-mutual information to characterize the fundamental limits under distortion and perception constraints. For scalar Gaussian sources, we derive closed-form expressions for the Rényi RDP function, showing that the perception constraint induces a feasible interval for the reproduction variance. Furthermore, we establish a Rényi-generalized version of the Strong Functional Representation Lemma. Our analysis reveals a phase transition in the complexity of optimal functional representations: for $0.5<α< 1$, the coding cost is bounded by the $α$-divergence of order $α+1$, necessitating a codebook with heavy-tailed polynomial decay; conversely, for $α> 1$, the representation collapses to one with finite support, offering new insights into the compression of shared randomness under generalized notions of mutual information.

Junle Li, Zhaoxiang Wu, Guanda Gong et al.

Liquid scintillator detectors are widely used in neutrino experiments due to their low energy threshold and high energy resolution. Despite the tiny abundance of $^{14}$C in LS, the photons induced by the $β$ decay of the $^{14}$C isotope inevitably contaminate the signal, degrading the energy resolution. In this work, we propose three models to tag $^{14}$C photon hits in $e^+$ events with $^{14}$C pile-up, thereby suppressing its impact on the energy resolution at the hit level: a gated spatiotemporal graph neural network and two Transformer-based models with scalar and vector charge encoding. For a simulation dataset in which each event contains one $^{14}$C and one $e^+$ with kinetic energy below 5 MeV, the models achieve $^{14}$C recall rates of 25%-48% while maintaining $e^+$ to $^{14}$C misidentification below 1%, leading to a large improvement in the resolution of total charge for events where $e^+$ and $^{14}$C photon hits strongly overlap in space and time.

Yuteng Zhang, Huifang Ma, Jiahui Wei et al.

Software vulnerability detection is critical for ensuring software security and reliability. Despite recent advances in deep learning, real-world vulnerability datasets suffer from two severe challenges: frequency imbalance and difficulty imbalance. We reinterpret these challenges from an embedding geometry perspective, observing that such imbalances induce geometric distortions in hyperspherical representation space. To address this issue, we propose MARGIN, a metric-based framework that learns discriminative vulnerability representations through adaptive margin metric learning and hyperspherical prototype modeling. MARGIN dynamically adjusts geometric regularization according to the distribution structure estimated by the von Mises-Fisher concentration, aligning the probability mass of embedding distributions with their corresponding Voronoi cells, thereby reducing geometric distortion and yielding more stable decision boundaries. Extensive experiments on public vulnerability datasets show that MARGIN consistently outperforms strong baselines, achieving notable improvements in classification and detection, especially on challenging, imbalanced datasets. Further analysis demonstrates that MARGIN produces more structured embedding geometries, improving robustness, interpretability, and generalization.

Jiahui Wei, Marios Kountouris

We introduce a Rényi Rate-Distortion-Perception-Privacy (R-RDPP) framework for indirect source coding. A latent source~$S$ is correlated with a private attribute~$U$, and the encoder observes only a noisy view~$X$ such that $(S,U) - X - Y$ holds at the decoder output~$Y$. The communication cost is measured by Sibson's $α$-mutual information $\Ialp$, the privacy leakage by $\Ibeta$, the semantic distortion between $S$ and $Y$, and the realism constraint at the semantic marginal $P_S$. We characterize the scalar Gaussian RDPP tradeoff, revealing that standard privacy metrics inherently penalize legitimate semantic recovery. To resolve this, we introduce a conditional privacy measure that quantifies only the residual leakage. In addition, we refine the achievability bounds for $α> 1$ via the Poisson functional representation. By deriving the exact geometric-mixture distribution of the Poisson index, we obtain exact closed-form expressions for integer-order Rényi entropies and sharper computable bounds in regimes where the resulting expression improves the logarithmic-moment approach.

Siyang Qin, Jiahui Wei, Roberto Manduchi

We introduce a new system for automatic image content removal and inpainting. Unlike traditional inpainting algorithms, which require advance knowledge of the region to be filled in, our system automatically detects the area to be removed and infilled. Region segmentation and inpainting are performed jointly in a single pass. In this way, potential segmentation errors are more naturally alleviated by the inpainting module. The system is implemented as an encoder-decoder architecture, with two decoder branches, one tasked with segmentation of the foreground region, the other with inpainting. The encoder and the two decoder branches are linked via neglect nodes, which guide the inpainting process in selecting which areas need reconstruction. The whole model is trained using a conditional GAN strategy. Comparative experiments show that our algorithm outperforms state-of-the-art inpainting techniques (which, unlike our system, do not segment the input image and thus must be aided by an external segmentation module.)