arXiv subjects

Noman Ali

Publications and source records attributed to Noman Ali.

3 recordsLinked to original sources

arXiv2026

Is Meta-Path Attention an Explanation? Evidence of Alignment and Decoupling in Heterogeneous GNNs

Meta-path-based heterogeneous graph neural networks aggregate over meta-path-induced views, and their semantic-level attention over meta-path channels is widely used as a narrative for ``which semantics matter.'' We study this assumption empirically by asking: when does meta-path attention reflect meta-path importance, and when can it decouple? A key challenge is that most post-hoc GNN explainers are designed for homogeneous graphs, and naive adaptations to heterogeneous neighborhoods can mix semantics and confound perturbations. To enable a controlled empirical analysis, we introduce MetaXplain, a meta-path-aware post-hoc explanation protocol that applies existing explainers in the native meta-path view domain via (i) view-factorized explanations, (ii) schema-valid channel-wise perturbations, and (iii) fusion-aware attribution, without modifying the underlying predictor. We benchmark representative gradient-, perturbation-, and Shapley-style explainers on ACM, DBLP, and IMDB with HAN and HAN-GCN, comparing against xPath and type-matched random baselines under standard faithfulness metrics. To quantify attention reliability, we propose Meta-Path Attention--Explanation Alignment (MP-AEA), which measures rank correlation between learned attention weights and explanation-derived meta-path contribution scores across random runs. Our results show that meta-path-aware explanations typically outperform random controls, while MP-AEA reveals both high-alignment and statistically significant decoupling regimes depending on the dataset and backbone; moreover, retraining on explanation-induced subgraphs often preserves, and in some noisy regimes improves, predictive performance, suggesting an explanation-as-denoising effect.

cs.LG↗

arXiv2025

Split-Fuse-Transport: Annotation-Free Saliency via Dual Clustering and Optimal Transport Alignment

Salient object detection (SOD) aims to segment visually prominent regions in images and serves as a foundational task for various computer vision applications. We posit that SOD can now reach near-supervised accuracy without a single pixel-level label, but only when reliable pseudo-masks are available. We revisit the prototype-based line of work and make two key observations. First, boundary pixels and interior pixels obey markedly different geometry; second, the global consistency enforced by optimal transport (OT) is underutilized if prototype quality is weak. To address this, we introduce POTNet, an adaptation of Prototypical Optimal Transport that replaces POT's single k-means step with an entropy-guided dual-clustering head: high-entropy pixels are organized by spectral clustering, low-entropy pixels by k-means, and the two prototype sets are subsequently aligned by OT. This split-fuse-transport design yields sharper, part-aware pseudo-masks in a single forward pass, without handcrafted priors. Those masks supervise a standard MaskFormer-style encoder-decoder, giving rise to AutoSOD, an end-to-end unsupervised SOD pipeline that eliminates SelfMask's offline voting yet improves both accuracy and training efficiency. Extensive experiments on five benchmarks show that AutoSOD outperforms unsupervised methods by up to 26% and weakly supervised methods by up to 36% in F-measure, further narrowing the gap to fully supervised models.

cs.CV↗

arXiv2024

Deep Fuzzy Framework for Emotion Recognition using EEG Signals and Emotion Representation in Type-2 Fuzzy VAD Space

Recently, the representation of emotions in the Valence, Arousal and Dominance (VAD) space has drawn enough attention. However, the complex nature of emotions and the subjective biases in self-reported values of VAD make the emotion model too specific to a particular experiment. This study aims to develop a generic model representing emotions using a fuzzy VAD space and improve emotion recognition by utilizing this representation. We partitioned the crisp VAD space into a fuzzy VAD space using low, medium and high type-2 fuzzy dimensions to represent emotions. A framework that integrates fuzzy VAD space with EEG data has been developed to recognize emotions. The EEG features were extracted using spatial and temporal feature vectors from time-frequency spectrograms, while the subject-reported values of VAD were also considered. The study was conducted on the DENS dataset, which includes a wide range of twenty-four emotions, along with EEG data and subjective ratings. The study was validated using various deep fuzzy framework models based on type-2 fuzzy representation, cuboid probabilistic lattice representation and unsupervised fuzzy emotion clusters. These models resulted in emotion recognition accuracy of 96.09\%, 95.75\% and 95.31\%, respectively, for the classes of 24 emotions. The study also included an ablation study, one with crisp VAD space and the other without VAD space. The result with crisp VAD space performed better, while the deep fuzzy framework outperformed both models. The model was extended to predict cross-subject cases of emotions, and the results with 78.37\% accuracy are promising, proving the generality of our model. The generic nature of the developed model, along with its successful cross-subject predictions, gives direction for real-world applications in the areas such as affective computing, human-computer interaction, and mental health monitoring.

cs.HC↗