SLONN: A Simulation Language for modeling of Neural Networks

Abstract

This paper presents a general purpose Simulation Language for modeling Of Neural Networks (SLONN) which has been implemented in our laboratory. Based on a new neuron model, SLONN can represent both spatial and temporal summation of a single neuron and synaptic plasticity. By introducing fork to describe a connection pattern between neurons and by using repetition connection, module type and module array to specify large networks, SLONN can be used to specify both small and large neural networks effectively. This language is distinguished by its hierarchical organization, which makes it possible to catch very general aspects at higher levels as well as very specific properties at lower levels. As an example to demonstrate some features of SLONN, we have modeled the habituation and sensitization behaviors in Aplysia. McClelland and Rumelhart, 1986;Lippmann, 1987). Although behaviors of neural networks are diverse, three things are crucial: a neuron model which serves as a computing unit; a connection pattern which connects the computing units to form a specific network; and a learning rule which is used for modifying connection weights in order to update network behaviors in a certain way.In order to facilitate the work of constructing neural networks and carrying out simulation experiments on a computer, i.e., to provide an easy environment for modelers of neural networks, we designed and implemented a highlevel, general purpose simulation language,

Using habituation as a simple and fundamental learning mechanism in an embodied artificial agent

“…Our models are designed at the abstract level of the IDSM. In this, they are different from most of the related work cited in the introduction (Bi et al, 2017; Hong et al, 2020; Marsland et al, 1999; Wang & Hsu, 1990) that work on the level of neurons or synapses. One slight advantage is that our models, being more conceptual, can be easier to analyze and understand, but, as discussed now, an in depth understanding of our models can still be complex.…”

“…Our models are designed at the abstract level of the IDSM. In this, they are different from most of the related work cited in the introduction (Bi et al, 2017;Hong et al, 2020;Marsland et al, 1999;Wang & Hsu, 1990) that work on the level of neurons or synapses. One slight advantage is that Figure 13.…”

“…This approach has for example been explored in (Marsland et al, 1999). In their paper, the authors consider both a simple linear integrator to modulate the synaptic efficacy of their modeled neurons and also a non-linear model introduced by (Wang & Hsu, 1990) which can account for long-term habituation effects. Their neuronal model of habituation is applied on a mobile robot and habituation is observed in the robot behavior: it eventually stops responding to repeated stimuli.…”

Improving Network Training on Resource-Constrained Devices via Habituation Normalization

“…Some researchers have established mathematical models of the habituation effects on the efficacy of a synapse, including Groves and Thompson [23], Stanley [24], and Wang and Hsu [25]. The model proposed by the Wang and Hsu considered a long-term memory, where the long-term memory means that an animal habituates more quickly to a stimulus to which it has been previously habituated.…”