Search for: All records

This study proposes a novel dynamical mechanism for pattern recognition discovered by interpreting a recurrent neural network (RNN) trained on a simple task inspired by the SET card game. We interpreted the trained RNN as recognizing patterns via phase shifts in a low-dimensional limit cycle in a manner analogous to transitions in a finite state automaton (FSA). We further validated this interpretation by handcrafting a simple oscillatory model that reproduces the dynamics of the trained RNN. Our findings not only suggest of a potential dynamical mechanism capable of pattern recognition, but also suggest of a potential neural implementation of FSA. Above all, this work contributes to the growing discourse on deep learning model interpretability. 

Surface adsorption of two commonly detected emerging contaminants, amlodipine (AMP) and carbamazepine (CBZ), onto model colloidal microplastics, natural organic matter (NOM), and fullerene nanomaterials have been investigated. It is found that AMP accumulation at these colloidal–aqueous interfaces is markedly higher than that of CBZ. Measurements of surface excess and particle zeta potential, along with pH-dependent adsorption studies, reveal a distinct influence of colloidal functional group on the adsorption properties of these pharmaceuticals. AMP shows a clear preference for a surface containing carboxylic group compared to an amine modified surface. CBZ, in contrast, exhibit a pH-dependent surface proclivity for both of these microparticles. The type of interactions and molecular differences with respect to structural rigidity and charge properties explain these observed behaviors. In this work, we also demonstrate a facile approach in fabricating uniform microspheres coated with NOM and C 60 nanoclusters. Subsequent binding studies on these surfaces show considerable adsorption on the NOM surface but a minimal uptake of CBZ by C 60 . Adsorption induced colloidal aggregation was not observed. These findings map out the extent of contaminant removal by colloids of different surface properties available in the aquatic environment. The methodology developed for the adsorption study also opens up the possibility for further investigations into colloidal–contaminant interactions.