Spontaneous generation of face recognition in untrained deep neural networks

Face-selective neurons are observed in the primate visual pathway and are considered the basis of facial recognition in the brain. However, it is debated whether this neuronal selectivity can arise spontaneously, or requires training from visual experience. Here, we show that face-selective neurons arise spontaneously in random feedforward networks in the absence of learning. Using biologically inspired deep neural networks, we found that face-selective neurons arise under three different network conditions: one trained using non-face natural images, one randomized after being trained, and one never trained. We confirmed that spontaneously emerged face-selective neurons show the biological view-point-invariant characteristics observed in monkeys. Such neurons suddenly vanished when feedforward weight variation declined to a certain level. Our results suggest that innate face-selectivity originates from statistical variation of the feedforward projections in hierarchical neural networks.

A total cost minimization control for wastewater treatment process by using extremum seeking control

This paper presents a real-time optimal control scheme based on a total cost index (TCI) by using the so-called extremum seeking control (ESC). The ESC searches the minimum value of a given performance index where the TCI consisting of the operational cost and the effluent quality cost is used as the index. An improved ESC is proposed where a transformed TCI by monotonically increasing function is utilized to improve the convergence property of the ESC while keeping the argument of the optimal point invariant. The feasibility of the ESC is tested for two types of pseudo-anaerobic-oxic process control: one is the return sludge recycle ratio control, and the other is the aeration control for an alternate zone which can be anaerobic or aerobic depending on the influent condition. Simulation study illustrates that the proposed ESC is able to find a near optimal point successfully and the TCI can be reduced by about 2.7 to 3.8% compared with that of a typical operating condition.