Future Work

Nai Ding’s recent and future work

Contents Topic 1: KIII’s effect on structure of data 1 KIII’s effect when trained by two points 2 KIII’s effect on data with different distributions Topic 2: Habituation’s impact on KIII

Nai Ding’s recent and future work

Topic 1: KIII’s effect on structure of data KIII simulate the mapping function of olfactory system, but what effect this map will have on data is still unknown and is the focus of this work.

1.KIII’s effect when trained by two points

Fig.1

If the KIII model is trained by two points, for example the big blue and red stars in fig.1, then the trained KIII model will map other points in the space to where? If other points, the small blue and red stars are mapped to a place pointed by the arrow initiate from the stars as in fig.1, then KIII

Nai Ding’s recent and future work

model’s effect is positive.

KIII is a stochastic network, perhaps with more training times the moving tendency of a specific point will by like fig.2.

Fig.2

Nai Ding’s recent and future work

2.KIII’s effect on data with different distributions

Fig.3

Nai Ding’s recent and future work

Topic 2: Habituation’s impact on KIII The KIII model also indicates that the response energy during habituation to undesired input is not decreased, but is shifted from the gamma range to the theta range of cortical activity. (“Tutorial on neurobiology: from single neurons to brain chaos”, W.J. Freeman)

I want to examine the changes in KIII when habituation happens.

Fig.4 Changes in frequency of oscillation at OB, when habituation happens.

Nai Ding’s recent and future work

Fig.4 Changes in weights between M notes in KIII model, when habituation happens.