Computational intelligence techniques have been used in wide applications. Out of numerous computational intelligence techniques, neural networks and support vector machines (SVMs) have been playing the dominant roles. However, it is known that both neural networks and SVMs face some challenging issues such as:(1) slow learning speed, (2) trivial human intervene, and/or (3) poor computational scalability. Extreme learning machine (ELM) as emergent technology which overcomes some challenges faced by other techniques has recently attracted the attention from more and more researchers. ELM works for generalized single-hidden layer feedforward networks (SLFNs). The essence of ELM is that the hidden layer of SLFNs need not be tuned. Compared with those traditional computational intelligence techniques, ELM provides better generalization performance at a much faster learning speed and with least human intervene. This paper gives a survey on ELM and its variants, especially on (1) batch learning mode of ELM, (2) fully complex ELM, (3) online sequential ELM, (4) incremental ELM, and (5) ensemble of ELM.

No abstract

Extreme learning machines (ELMs) have been proposed for generalized single-hidden-layer feedforward networks which need not be neuron-like and perform well in both regression and classification applications. In this brief, we propose an ELM with adaptive growth of hidden nodes (AG-ELM), which provides a new approach for the automated design of networks. Different from other incremental ELMs (I-ELMs) whose existing hidden nodes are frozen when the new hidden nodes are added one by one, in AG-ELM the number of hidden nodes is determined in an adaptive way in the sense that the existing networks may be replaced by newly generated networks which have fewer hidden nodes and better generalization performance. We then prove that such an AG-ELM using Lebesgue p-integrable hidden activation functions can approximate any Lebesgue p-integrable function on a compact input set. Simulation results demonstrate and verify that this new approach can achieve a more compact network architecture than the I-ELM.

No abstract

No abstract

No abstract

Actin, a highly conserved protein comprising cell stress fibers and other cellular structures, is found in both the cytoplasm and nucleus of cells and responds to both epigenetic signals and altered gene expression occurring during tumorigenesis. We have previously shown that changes in the cytoplasmic F-and G-actin ratios reflect bladder cancer risk. To determine whether nuclear actin is also altered and how nuclear and cytoplasmic actin alterations are interrelated in transformation, an in vitro model of carcinogen-induced transformation consisting of 2 human uroepithelial cell lines immortalized by infection with SV-40 was studied. One line, HUC-PC, is tumorigenic in nude mice after incubation with the carcinogen 4-ABP, the other, HUC-BC, is not. Cytoplasmic and nuclear F-and G-actin were determined by QFIA on individual cells using fluorochrome-labeled phallicidin and DNase, I, respectively. Before exposure to 4-ABP, the PC cells had lower cytoplasmic F-actin content, higher cytoplasmic G-actin content, but similar levels of nuclear G-and Carcinogenesis is a multi-stage process resulting in the eventual emergence of a malignant phenotype that often continues to progress to an eventually fatal metastatic phenotype. Although the genetic elements of carcinogenesis, the oncogenes and tumor suppressor genes, have been considered the central theme and are relatively well understood (Bishop, 1987;Weinberg, 1991), it is clear that cancer is a more complex process also involving epigenetic interactions among cells and their environment. Rubin (1992) described reversible transformation, including stable genetic heterogeneity of the cell population, simply by manipulating cell culture growth conditions. Pienta et al. (1989) have proposed that DNA organization in nuclei is an important epigenetic factor in cancer development, whereas others have observed that cell-cell interactions such as gap junctions and adherence are also involved (Trosko et al., 1990). Other experiments studying the relationship between basement membrane and type IV procollagen mRNA expression support a direct interaction among extracellular matrix/ cytoskeleton/nuclear matrix (Schlussel et al., 1991). Other studies show a connection between cytoskeletal actin, the nucleus, and the cell periphery that is altered after transformation (Pienta and Coffey, 1992). Mutation of the b-actin gene has been shown to alter the metastatic phenotype (Leavitt, 1994). A number of genes are also known to regulate the multimolecular focal complexes associated with actin stress fibers and cell cycle progression through G 1 and DNA synthesis (Nobes and Hall, 1995;Olson et al., 1995). One candidate for being the crucial integrative factor is the actincontaining nuclear matrix.Qualitative alterations of cytoplasmic actin, the major cellular protein involved in cell-cell interaction, morphology, motility, differentiation and adhesion, have been widely investigated in cancer cells in response to interactions between cells and their environment (Rao and Cohen, 1991;Pollard...

Extreme learning machines (ELMs) have been proposed for generalized single-hidden-layer feedforward networks which need not be neuron alike and perform well in both regression and classification applications. The problem of determining the suitable network architectures is recognized to be crucial in the successful application of ELMs. This paper first proposes a dynamic ELM (D-ELM) where the hidden nodes can be recruited or deleted dynamically according to their significance to network performance, so that not only the parameters can be adjusted but also the architecture can be self-adapted simultaneously. Then, this paper proves in theory that such D-ELM using Lebesgue p-integrable hidden activation functions can approximate any Lebesgue p-integrable function on a compact input set. Simulation results obtained over various test problems demonstrate and verify that the proposed D-ELM does a good job reducing the network size while preserving good generalization performance.