Higher order data compression with neural networks

Namphol, A.; Arozullah, M.; Chin, Steven H.

doi:10.1109/ijcnn.1991.155149

Cited by 20 publications

(3 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A neural network model which reflects the current trend in structured network design has been proposed in [26] for image compression applications. The structure of this network (shown in Fig.…”

Section: Implementing a Block-connected Multilayer Neural Networkmentioning

confidence: 99%

See 1 more Smart Citation

Implementing regularly structured neural networks on the DREAM machine

Shams

Gaudiot

1995

IEEE Trans. Neural Netw.

View full text Add to dashboard Cite

High-throughput implementations of neural network models are required to transfer the technology from small prototype research problems into large-scale "real-world" applications. The flexibility of these implementations in accommodating for modifications to the neural network computation and structure is of paramount importance. The performance of many implementation methods today is greatly dependent on the density and the interconnection structure of the neural network model being implemented. A principal contribution of this paper is to demonstrate an implementation method which exploits maximum amount of parallelism from neural computation, without enforcing stringent conditions on the neural network interconnection structure, to achieve this high implementation efficiency. We propose a new reconfigurable parallel processing architecture, the Dynamically Reconfigurable Extended Array Multiprocessor (DREAM) machine, and an associated mapping method for implementing neural networks with regular interconnection structures. Details of the system execution rate calculation as a function of the neural network structure are presented. Several example neural network structures are used to demonstrate the efficiency of our mapping method and the DREAM machine architecture on implementing diverse interconnection structures. We show that due to the reconfigurable nature of the DREAM machine, most of the available parallelism of neural networks can be efficiently exploited.

show abstract

Section: Implementing a Block-connected Multilayer Neural Networkmentioning

confidence: 99%

“…sion with a regularly blocked structure[26].A neural network structure for image compression and decompres DREAM Machine implementation vs. 52% for the fixed ring implementation.…”

mentioning

confidence: 99%

Implementing regularly structured neural networks on the DREAM machine

Shams

Gaudiot

1995

IEEE Trans. Neural Netw.

View full text Add to dashboard Cite

show abstract

“…Another variation proposed by Namphod et al [51] involves five layers and two sets of training. Initially, the outer two layers are trained with the inner three layers modeled as one layer.…”

Section: F Autoassociative Transform Codingmentioning

confidence: 99%

Neural network approaches to image compression

1995

View full text Add to dashboard Cite

This paper presents a tutorial overview of neural networks as signal processing tools for image compression. They are well suited to the problem of image compression due to their massively parallel and distributed architecture. Their characteristics are analogous to some of the features of our own visual system, which allow us to process visual information with much ease. For example, multilayer perceptrons can be used as nonlinear predictors in differential pulse-code modulation (DPCM). Such predictors have been shown to increase the predictive gain relative to a linear predictor. Another active area of research is in the application of Hebbian learning to the extraction of principal components, which are the basis vectors for the optimal linear Karhunen-Loève transform (KLT). These learning algorithms are iterative, have some computational advantages over standard eigendecomposition techniques, and can be made to adapt to changes in the input signal. Yet another model, the self-organizing feature map (SOFM), has been used with a great deal of success in the design of codebooks for vector quantization (VQ). The resulting codebooks are less sensitive to initial conditions than the standard LBG algorithm, and the topological ordering of the entries can be exploited to further increase coding efficiency and reduce computational complexity.

show abstract

Traffic and video quality with adaptive neural compression

et al. 1996

View full text Add to dashboard Cite

Video sequences are major sources of traffic for broadband ISDN networks, and video compression is fundamental to the efficient use of such networks. We present a novel neural method to achieve real-time adaptive compression of video. This tends to maintain a target quality of the decompressed image specified by the user. The method uses a set of compression/decompression neural networks of different levels of compression, as well as a simple motiondetection procedure. We describe the method and present experimental data concerning its performance and traffic characteristics with real video sequences. The impact of this compression method on ATM-cell traffic is also investigated and measurement data are provided.

show abstract

Higher order data compression with neural networks

Cited by 20 publications

References 1 publication

Implementing regularly structured neural networks on the DREAM machine

Implementing regularly structured neural networks on the DREAM machine

Neural network approaches to image compression

Traffic and video quality with adaptive neural compression

Contact Info

Product

Resources

About