Active learning methods aim to improve sample complexity in machine learning. In this work, we investigate an active learning scheme via a novel gradient-free cutting-plane training method for ReLU networks of arbitrary depth and develop a convergence theory. We demonstrate, for the first time, that cutting-plane algorithms, traditionally used in linear models, can be extended to deep neural networks despite their nonconvexity and nonlinear decision boundaries. Moreover, this training method induces the first deep active learning scheme known to achieve convergence guarantees, revealing a geometric contraction rate of the feasible set. We exemplify the effectiveness of our proposed active learning method against popular deep active learning baselines via both synthetic data experiments and sentimental classification task on real datasets.

The increasing challenges caused by contaminated soil and frequent climate changes have significantly impacted traditional agricultural systems. Indoor aquaponic systems, which utilize nutrient-rich fish tank water to cultivate plants, present a viable solution by eliminating the need for soil and providing a controlled environment that mitigates the effects of climate change. However, maintaining a balanced nutrient composition for fish and plants remains challenging. Integrating the Internet of Things (IoT) into aquaponic systems enables realtime monitoring of nutrient levels, ensuring optimal system maintenance. In this paper, we present the development of a portable and low-cost Aquaponic Monitoring System designed to automate nutrient data collection, reducing the need for manual labor while enhancing system efficiency. This system not only addresses the current challenges but also opens up exciting possibilities for further optimization, ultimately increasing yield and sustainability in aquaponic farming.