Abstract
Recurrent neural networks have the potential to develop internal representations that allow useful encoding of the dynamics behind a sequence of inputs. In this paper we present a model of time-varying prabability distributions by using a two-layered columnar recurrent neural network in which each hidden unit has recurrent connections from the context units representing delayed outputs of the hidden unit. The probability distribution model can provide predictions in terms of a given probabilistic density function of the context units instead of the single guess which is usually provided by Elman-type recurrent neural networks. The advantage of this approach is the interpretability between the context units and the dynamics behind the inputs. Computer simulations of a stochastic grammar and a discrete trend time-sequence are shown to demonstrate the capability of the model.
| Original language | English |
|---|---|
| Pages (from-to) | 183-191 |
| Number of pages | 9 |
| Journal | Fujitsu Scientific and Technical Journal |
| Volume | 32 |
| Issue number | 2 |
| Publication status | Published - 1996 |
| Externally published | Yes |
Fingerprint
ASJC Scopus subject areas
- Electrical and Electronic Engineering
Cite this
Columnar recurrent neural network and time series analysis. / Matsuoka, Masahiro; Golea, Mostefa; Sakakibara, Yasubumi.
In: Fujitsu Scientific and Technical Journal, Vol. 32, No. 2, 1996, p. 183-191.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - Columnar recurrent neural network and time series analysis
AU - Matsuoka, Masahiro
AU - Golea, Mostefa
AU - Sakakibara, Yasubumi
PY - 1996
Y1 - 1996
N2 - Recurrent neural networks have the potential to develop internal representations that allow useful encoding of the dynamics behind a sequence of inputs. In this paper we present a model of time-varying prabability distributions by using a two-layered columnar recurrent neural network in which each hidden unit has recurrent connections from the context units representing delayed outputs of the hidden unit. The probability distribution model can provide predictions in terms of a given probabilistic density function of the context units instead of the single guess which is usually provided by Elman-type recurrent neural networks. The advantage of this approach is the interpretability between the context units and the dynamics behind the inputs. Computer simulations of a stochastic grammar and a discrete trend time-sequence are shown to demonstrate the capability of the model.
AB - Recurrent neural networks have the potential to develop internal representations that allow useful encoding of the dynamics behind a sequence of inputs. In this paper we present a model of time-varying prabability distributions by using a two-layered columnar recurrent neural network in which each hidden unit has recurrent connections from the context units representing delayed outputs of the hidden unit. The probability distribution model can provide predictions in terms of a given probabilistic density function of the context units instead of the single guess which is usually provided by Elman-type recurrent neural networks. The advantage of this approach is the interpretability between the context units and the dynamics behind the inputs. Computer simulations of a stochastic grammar and a discrete trend time-sequence are shown to demonstrate the capability of the model.
UR - http://www.scopus.com/inward/record.url?scp=0030398085&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0030398085&partnerID=8YFLogxK
M3 - Article
AN - SCOPUS:0030398085
VL - 32
SP - 183
EP - 191
JO - Fujitsu Scientific and Technical Journal
JF - Fujitsu Scientific and Technical Journal
SN - 0016-2523
IS - 2
ER -