Search for: All records

Soil microbial fuel cells (SMFCs) function as bioelectrochemical energy harvesters that convert electrons stored in soil organic matter into useful electrical energy. Broadly, an SMFC comprises three essential components: an anode buried in the soil (the negative terminal), a colony of exoelectrogenic microorganisms residing on this anode, and a cathode (the positive terminal). As the exoelectrogens respire, they release electrons to the anode, which acts as an external receptor. These released electrons then flow through a load (e.g. a resistor), connecting the anode and cathode. Though minuscule, the electrical power produced by SMFCs has a number of potential applications such as sustaining low-power embedded electronics, pollutant remediation, or as a bio-sensing proxy for soil qualities and microbial activity. This discussion aims to emphasize the potential of SMFCs in addressing real-world environmental issues and to generate interest in the larger scientific community for broad interdisciplinary research efforts, particularly in field deployments. 

We present REGLO, a novel methodology for repairing pretrained neural networks to satisfy global robustness and individual fairness properties. A neural network is said to be globally robust with respect to a given input region if and only if all the input points in the region are locally robust. This notion of global robustness also captures the notion of individual fairness as a special case. We prove that any counterexample to a global robustness property must exhibit a corresponding large gradient. For ReLU networks, this result allows us to efficiently identify the linear regions that violate a given global robustness property. By formulating and solving a suitable robust convex optimization problem, REGLO then computes a minimal weight change that will provably repair these violating linear regions.