Fully-connected Committee Machine (FCM) based Online Leaning under Concept Drift
K Prasanna · Zenodo (CERN European Organization for Nuclear Research) · 2021
The term “drift” refers to unanticipated changes in the transmission of data in the primary distribution over time. Conceptual drift research entails developing methods and strategies for detecting, interpreting, and adapting to drift. Machine learning approaches can produce poor learning outcomes in the conceptual drift environment if drift is not addressed. Furthermore, due to developments in the concept of drift, revealing a method not mentioned in the literature, the concept of drift learning methodologies has been significantly systematic in recent years. We used a layered neural network framework to experiment with different scenarios of online learning under concept drift using a fully-connected committee machine (FCM). We conduct experiments in various scenarios using a layered neural network framework for online learning under concept drift. In neural layered networks, sigmoidal and ReLU activation functions are considered for learning regression. When the layered framework is trained from the input dynamic data stream, the regression scheme changes consciously in all scenarios. A fully-connected committee machine (FCM) is trained to perform the tasks described in online learning with M hidden units on dynamically generated inputs. In this method, we run Monte Carlo simulations with the same number of units on both sides, K and M, to define the dynamic advancement of intersections between several hidden units and the calculation of generalization error. This is applied to overlearnability as a method of over-forgetting, integrating weight decay, and examining its effects in the presence of concept drift.