Fair graph clustering seeks partitions that respect network structure while maintaining proportional representation across sensitive groups, with applications spanning community detection, team formation, resource allocation, and social network analysis. Many existing approaches enforce rigid constraints or rely on multi-stage pipelines (e.g., spectral embedding followed by $k$-means), limiting trade-off control, interpretability, and scalability. We introduce \emph{DFNMF}, an end-to-end deep nonnegative tri-factorization tailored to graphs that directly optimizes cluster assignments with a soft statistical-parity regularizer. A single parameter $λ$ tunes the fairness--utility balance, while nonnegativity yields parts-based factors and transparent soft memberships. The optimization uses sparse-friendly alternating updates and scales near-linearly with the number of edges. Across synthetic and real networks, DFNMF achieves substantially higher group balance at comparable modularity, often dominating state-of-the-art baselines on the Pareto front. The code is available at https://github.com/SiamakGhodsi/DFNMF.git.

Abstract Group work is a prevalent activity in educational settings, where students are often divided into topic-specific groups based on their preferences. The grouping should reflect students’ aspirations as much as possible. Usually, the resulting groups should also be balanced in terms of protected attributes like gender, as studies suggest that students may learn better in mixed-gender groups. Moreover, to allow a fair workload across the groups, the cardinalities of the different groups should be balanced. In this paper, we introduce a multi-fair capacitated (MFC) grouping problem that fairly partitions students into non-overlapping groups while ensuring balanced group cardinalities (with a lower and an upper bound), and maximizing the diversity of members regarding the protected attribute. To obtain the MFC grouping, we propose three approaches: a greedy heuristic approach, a knapsack-based approach using vanilla maximal knapsack formulation, and an MFC knapsack approach based on group fairness knapsack formulation. Experimental results on a real dataset and a semi-synthetic dataset show that our proposed methods can satisfy students’ preferences and deliver balanced and diverse groups regarding cardinality and the protected attribute, respectively.

Nowadays, urban mobility plays an important role in modern cities for city planning, navigation, and other mobility services. Taxicabs are vital public services in large cities that are taken by passengers thousands of times every day. Reducing the number of vacant vehicles on the streets will help service providers to raise drivers' incomes, reduce energy consumption, optimize traffic efficiency, and control air pollution problems in large cities. Since drivers do not have enough information about the location of passengers and other taxis, most of them might drive to the same area. Due to the lack of passenger information, they often end up without picking up any passengers while there are highly demanded areas in their neighborhood. To address these issues, machine learning techniques can be applied to analyze mobility data acquired from the IoT sensors and help companies to organize the taxi fleet or minimize the wait-time for both passengers and drivers in the city. In this paper, an LSTM-based deep sequence learning model is applied to forecast taxi-demand in a particular urban area in a smart city. For this purpose, points of interest (POIs) in the city are extracted from Google Maps and integrated with the mobility data sources. Given a real-world dataset and two evaluation metrics, we observed that taxi-demand in each urban area can be influenced by external factors such as neighborhood locations and the POIs located in that area. The results show that the proposed method outperforms the vanilla LSTM model and has less average error than baseline methods in terms of the Mean Squared Error (MSE) and Symmetric Mean Absolute Percentage Error (SMAPE).