OFDM Multiple Antennas Cognitive Radio Relay CDMA Synchronization Channel Estimation Spectrum Sharing Interference Cancellation Resource Allocation Spectrum Sensing Neural Networks Full duplex Stochastic Geometry Equalizer Bi-Directional Feedback Energy Harvesting Heterogeneous Networks Femtocell Device-to-Device (D2D) Idle cells Cross-link interference FBMC Spectral efficiency Cell Search SINR mismatch problem NOMA Ultra-dense small cell networks HetNet interference management Dynamic TDD outage probability selection diversity achievable sum rate bursty traffic model Cognitive relay networks mode selection multi-spectral 5G Complexity Singular Vale Decomposition OQAM tabu-search Filtered OFDM TDD configuration flexible duplex Handoff GFDM Heterogeneous channel estimation capability self-interference cancellation in-band full-duplex system Channel estimation error coexistence CP-OFDM MU-MIMO automatic repeat request (ARQ) Two-way communications UWB full-duplex relay full-duplex cellular Simultaneous Sensing and Transmission Correlated MIMO transmission capacity (TC) sensing duration Bi-directional full-duplex Vehicle-to-Vehicle prototype filter pilot signal Coexistence scenarios resource size control Vehicle-to-vehicle communication link reliability interference to noise ratio eigen decomposition TS-W-OFDM Resource management Cooperative systems LTE-based V2V Aggregate interference time-frequency efficiency mixed numerology Windowing Reliability C-V2V Asynchronous Transmission Full-duplex Computation offloading Grant-free Transmission Preamble full-spreading NOMA massive connectivity Edge computing Multiple access MLP Deep learning Railway Mobility interference mitigation HST non-orthogonal multiple access
Status : Published 
Date : 2004-10 
Title : Decision Boundary Formation of Neural Networks 
Authors : Chulhee Lee, E. Jung, O. Kwon, M. Park and Daesik Hong 
Journal : World Scientific and Engineering Academy and Society Transaction on Computers 
Abstract : In this paper, we provide a thorough analysis of decision boundaries of neural networks when they are
used as a classifier. First, we divide the classifying mechanism of the neural network into two parts: dimension
expansion by hidden neurons and linear decision boundary formation by output neurons. In this paradigm, the input
data is first warped into a higher dimensional space by the hidden neurons and the output neurons draw linear
decision boundaries in the expanded space (hidden neuron space). We also found that the decision boundaries in
the hidden neuron space are not completely independent. This dependency of decision boundaries is extended to
multiclass problems, providing a valuable insight into formation of decision boundaries in the hidden neuron space.
This analysis provides a new understanding of how neural networks construct complex decision boundaries and
explains how different sets of weights may provide similar results. 
List of Articles
No.
Statussort Date
» [Other Journals] Chulhee Lee, E. Jung, O. Kwon, M. Park and Daesik Hong, "Decision Boundary Formation of Neural Networks", WSEAS Trans. Computers, Oct 2004 Published  2004-10 
9 [Optical Engineering] Kyunggoo Lee, Sooyong Choi, Sunghwan Ong, Cheolwoo You, and Daesik Hong, "Equalization techniques using neural networks for digital versatile disk-read-only memory systems", Optical Eng., Feb 1999 Published  1999-02 
8 [IEEE Trans. Neural Netw.] Cheolwoo You and Daesik Hong, "Nonlinear Blind Equalization Schemes Using Complex-valued Multilayer Feedforward Neural Networks", IEEE Trans. Neural Networks, Nov 1998 Published  1998-11 
7 [IEEE Commun. Lett.] Sooyong Choi, Sunghwan Ong, Cheolwoo You and Daesik Hong, "A Quadratic Sigmoid Neural Equalizer for Nonlinear Digital Magnetic Recording Channels", IEEE Comm. Letters, Sep 1998 Published  1998-09 
6 [Neuro Computing] Sangmok Lee, Cheolwoo You and Daesik Hong, "Rapid Acquisition Using A Neural Network in DS/SS Communication System", Neurocomputing, Nov 1997 Published  1997-11 
5 [IEEE Trans. Signal Process.] Sunghwan Ong, Cheolwoo You, Sooyong Choi and Daesik Hong, "A Decision Feedback Recurrent Neural Equalizer as an Infinite Impulse Response Filter", IEEE Trans. Signal Processing, Nov 1997 Published  1997-11 
4 [IEEE Trans. Magn.] Sunghwan Ong, Sooyong Choi, Cheolwoo You, and Daesik Hong, "A Decision Feedback Recurrent Neural Equalizer for Digital Communication", IEEE Trans. Magnetics, Sep 1997 Published  1997-09 
3 [IEEE Trans. Magn.] Sooyong Choi, Sunghwan Ong, Cheolwoo You, Daesik Hong and Jachee Cho, "Performance of Neural Equalizers on Partial Erasure Model", IEEE Trans. Magnetics, Sep 1997 Published  1997-09 
2 [IEEE Trans. Ind. Electron.] Daesik Hong and Okan K. Ersoy, "Parallel, Self-Organizing Hierarchical Neural Networks-II", IEEE Trans. Industrial Elec., April 1993 Published  1993-04 
1 [IEEE Trans. Neural Netw.] Daesik Hong and Okan K. Ersoy, "Parallel, Self-Organizing Hierarchical Neural Networks", IEEE Trans. Neural Net., June 1990 Published  1990-06