Diploma and Master Theses (authored and supervised):
"Towards Distributed Controllers Based on Caenorhabditis elegans Locomotory Neural Network";
Supervisor: R. Grosu, R. M. Hasani;
Institut für Technische Informatik,
final examination: 2016-12-22.
In the present study, we aim to understand neuronal controlling mechanisms by investigat- ing the locomotory neural circuit of the nematode Caenorhabditis elegans (C. elegans). C. elegans is a transparent 1mm roundworm which naturally inhabits in soil. Its stereotypic nervous system consists of only 302 identifiable neurons hard-wired through approxi- mately 5000 chemical synapses and 2000 gap junctions. Because of highly concentrated biological research on its neuronal network, C. elegans is one of the promising models to learn the controlling and learning principles applicable in development of brain-inspired artificial intelligence. We are particularly interested in the part of the locomotory NC -
Tap Withdrawal (TW) neuronal circuit - responsible for the processing of the mechanical tap stimulus.
We applied synaptic and neuronal "computer" ablations to measure the impact of reducing neuronal structure on time spent on specific direction of locomotion and membrane potential of neurons. The minimal forward- and backward-responsible circuit have been constructed by adding connections from scratch. The minimal circuits are merged to check the modularity of two behavioral opposite circuits.
We have identified crucial chemical and electrical synapses controlling the forward and backward tap withdrawal. We have reduced them to find the minimal number of connections preserving the correct behavioral output. Neuronal ablations emphasize the premier role of the excitatory neurons within the neural circuit. The technique of building the minimal circuits from scratch allowed us to identify functional pathways and cycles responsible for (i) recognizing the start and end of the stimulus, (ii) behavioral concurrency and (iii) structural support for default forward locomotion. By putting two minimal circuits together, we identified overlapping parts of the circuits, crucial for both, anterior and posterior taps.
Finally, based on the acquired knowledge, we introduce a new fashion in designing of neuronal controllers by implementing simple stock market decision module. The decision module is composed of two sub-modules: 1. Indicator evaluation module compares the current and historical value of chosen stock market indicator, 2. C. elegans TW circuit mapping the forward and backward commands to BUY or SELL stocks.
Created from the Publication Database of the Vienna University of Technology.