University of Michigan–Dearborn

Smart agriculture requires efficient and scalable sensing of key soil parameters. While recent wireless sensing systems provide low-cost alternatives, active designs depend on powered underground components with potential instability and contamination risks, whereas passive battery-free approaches suffer from limited accuracy and are typically restricted to moisture detection. This work presents SoilNutri, a fully passive and battery-free metasurface capsule that enables simultaneous, high-precision monitoring of soil moisture and nitrogen concentration. The system employs a complementary split-ring (CSR) metasurface integrated within a sandwich-structured capsule that provides electromagnetic isolation and stable coupling with the surrounding soil. A lightweight two-stage soil-meta-VAE framework reconstructs soil-distorted reflection spectra and jointly predicts moisture and nitrogen from a shared latent representation. Extensive in-lab and field experiments demonstrate high-precision moisture estimation with 0.64% MAE and ppm-level nitrogen prediction with 5.14 ppm MAE across diverse soil types, depths, and seasonal conditions. With a per-unit cost below $1 and a reusable encapsulated design, SoilNutri offers a scalable, sustainable, and physically interpretable solution for multi-nutrient, in-situ soil monitoring for next-generation smart agriculture.

This article reconstructs and extends David Gordon’s distinction between capitalist and socialist efficiency as a framework for analyzing the co‑evolution of production techniques, class power, and worker subjectivity. Historical conflicts such as the Lordstown strike and new forms of algorithmic supervision in digital labor markets are used to illustrate the relevance of Gordon’s dual conception of efficiency. The article explores how workers resist these capitalist control strategies and how emerging practices—platform cooperatives, online worker forums, and community‑based ownership efforts—prefigure socialist forms of production by cultivating democratic capacities. It extends Gordon’s conception of socialist efficiency from the workplace to the political realm of democratic self‑governance. Finally, it argues that such a framework must be expanded to incorporate ecological constraints, by proposing an eco‑socialist conception of quantitative efficiency grounded in ecological sustainability and the development of human capabilities. Using Gordon’s article as the foundation, this framework offers a perspective for understanding transitions toward democratic, sustainable, post‑capitalist social formation. JEL Classification: B51, J52, J53, P3

In preparation for Issue 2, the Journal’s editorial board reached out to its network of organizers to invite short statements on their current work. We asked that these reflections address the guiding questions for this issue, including how their work intersects with social movements, unions or electoral politics, their experiences of working in and against authoritarianism, and their efforts to defend or promote democracy. The following statements from colleagues across three continents offer insights from diverse organizing networks and roles: an organizer turned local official in Hungary; a director of Faith in Action affiliates for Central America; an immigration organizer for the African diaspora in Detroit, and a lead organizer for Berlin, representing one of the largest German organizing networks.

Brain tumors remain a critical clinical challenge in which timely and accurate diagnosis strongly influences outcomes. Conventional convolutional neural network pipelines often demand substantial computational resources and large labeled datasets, limiting deployment in resource-constrained or real-time settings. In response, this study targets competitive multi-class magnetic resonance imaging classification with improved generalizability and computational efficiency through a lightweight pipeline that combines standardized preprocessing, handcrafted feature extraction (shape, intensity, and texture descriptors), correlation-based feature selection, and an optimally tuned ensemble of classical learners including Logistic Regression, k-Nearest Neighbors, Support Vector Machine, Decision Tree, Random Forest (RF), Gradient Boosting, AdaBoost, and Extreme Gradient Boosting (XGB) using Optuna for hyperparameter optimization. Evaluation employs accuracy, precision, recall, F1 score, specificity, Matthews Correlation Coefficient (MCC), Cohen’s Kappa, and the Area Under the Receiver Operating Characteristic Curve. Tuned XGB and RF establish strong single-model baselines (e.g., XGB: accuracy 0.9505, F1 score 0.9468), while a soft-voting ensemble attains accuracy 0.9824 and Area Under the Curve 0.99, with a high MCC and Cohen’s Kappa indicating robust, well-calibrated predictions. Overall, the findings suggest that the optimized classical ensemble with strategic choice of features can be as effective as deep learning and provide a lower computational cost as well as improved interpretability in clinical processes.

In studies of Alzheimer's disease (AD), limited sample size considerably hampers the performance of intelligent diagnostic systems. Using multi-site data increases sample size but raises concerns regarding data privacy and inter-site heterogeneity. To address these issues, we developed a knowledge-guided federated graph attention learning network with a diffusion module to facilitate AD diagnosis from multi-site data. We used multiple templates to extract regions-of-interest (ROI)-based volume features from structural magnetic resonance imaging (sMRI) data. These volume features were then combined with previously identified AD features from published studies (prior knowledge) to determine the discriminative features within the images. We then designed an attention-guided diffusion module to synthesize samples by prioritizing these key features. The diffusion module was trained within a federated learning framework, which ensured inter-site data privacy while limiting data heterogeneity. Finally, we designed a federated graph attention learning network as a classifier to capture AD-related deep features and improve the accuracy of diagnosing AD. The efficacy of our approach was validated using three AD datasets. Thus, the classifier developed in this study represents a promising tool for optimizing multi-site neuroimaging data to improving the accuracy of diagnosing AD in the clinic.