Associative Learning using a Deep neural network based on unsupervised representation

Learning is the way we acquire knowledge about the world around us, and it is through this process of knowledge acquisition, that the environment alerts our behavioural responses. The acquisition of knowledge is an active and ongoing cognitive process based on our perceptions. In the brain, knowledge is learnt by associating different sensory data types, such as image and voice.

Associative memory plays a crucial role in our ability to link things together and forms a large part of human intelligence. Associative memory models have previously been used to imitate such a learning process [3], but these simpler architectures have failed to deal with large complex data. The concept of associative memories matches well with the processes between stimulus associations.

Experiencing co-occurrent stimuli, i.e. Figure 1 lighting and thunder forms two kinds of links. These stimuli connect and form a representation of the lighting auto-associative (AA). The mind also learns another association lighting-thunder Hetero-associative (HA). Psychology theory distinguishes between these two processes and assigns them an order,

Perceptual mechanisms form the representation. Process of predictive learning follows, operating over those representations.