A Neural Network Model for a Hierarchical Spatio-temporal Memory

Face Detector Design for an Expression Robot

“…This correlation of features is also discussed in McClelland and Rogers (2003) who argue that information should be stored at the individual concept level rather than at the super ordinate category level allowing properties to be shared by many items. The Hubel Weisel model of conceptual memory (Ramanathan et al, 2010) is the first model that implements the bottom up hierarchical processing of Hubel and Weisel out of the domain of object recognition has attempted to model hierarchical representation of keywords to form concepts.…”

“…In this section we describe briefly the Hubel Weisel model of conceptual memory (Ramanathan et al, 2010) which is organized in a bottom up hierarchy. This means that the component features are represented lower in the hierarchy before the representation of concept objects.…”

“…While the algorithm was effective, loss of old memories still occurred, despite the introduction of pseudo vectors. The behavior of the model during incremental learning is further discussed in Ramanathan et al (2010). One aspect of memory representation in the model, however, was found curious and is discussed as below.…”

“…Prior work (Ramanathan et al, 2010) describes how a bottom up hierarchy of Growing Self Organizing maps (GSOM) (Alahakhoon et al, 2000) was created to explore bottom up learning in Hubel Weisel models specifically for representing a hierarchy of concepts. The paper showed that the Hubel Weisel model of concept memory captured certain cognitive properties of semantic cognition (Rogers and McClelland, 2003) such as hierarchical representation, multiple similarities (Sloutsky, 2003) and chain retrieval (Widrow, 2009).…”