The principles of “green chemistry” establish the definite demands for chemical procedures in carrying out synthetic or analytic methods using compounds extracted from biological resources. Thus, nanoparticle synthesis based on components extracted from bacteria, fungi, algae and plantae has been used for wide applications. Natural proteins, lipids, carbohydrates and unsaturated aliphatic (aromatic) compounds have the capability to act as reducing and stabilizing agents in nanoparticle synthesis. Gold nanoparticles (GNPs) have been known to be characterized by some specific features, viz. plasmonic ones, which are the main factors in the exposure of SPR, LPR and SERS. The formation of a limited area in isotropic and homogeneous media is a necessary condition for the appearance of a new phase (nanoreactor formation), which is the result of kinetic (diffusion) and spatial effects (thepresence of both structural and electrostatic obstacles). Taken together, these lead to some peculiarities in the region where nanostructure growth occurs. A nanoreactor can be viewed as a nanobot, defined as “a controllable nanoscale machine composed of a sensor and motor used to perform specific tasks specified by the appropriate conditions”. The specific task is GNP formation, the sensor is the environment condition (temperature and pH) and the motor (driven forces) is the specific medium conformation where the GNPs are formed. In order to recognize a nanobot's features, we determine the factors (environment content, its temperature and pH) influencing its capability to form GNPs. Such recognition is facilitated by the data obtained from TEM, SPR and SERS on the formed GNPs. The main aim of our investigation is to provide the method for nanobot feature recognition using a model of a multilayer fully connected perceptron whose architecture includes several hidden layers with different numbers of neurons to ensure the depth of the learning and the ability to process the complex dependencies in the data.