The application of neural networks to induction machine control

Neural networks (NNs) are discussed in connection with their possible use in induction machine drives. The mathematical model of the NN as well as a commonly used learning algorithm is presented. Possible applications of NNs to induction machine control are discussed. A simulation of an NN successfully identifying the nonlinear multivariable model of an induction-machine stator transfer function is presented. Previously published applications are discussed, and some possible future applications are proposed.

Identification and control of induction machines using artificial neural networks

The use of artificial neural networks (ANNs) to identify and control induction machines is proposed. Two systems are presented: a system to adaptively control the stator currents via identification of the electrical dynamics, and a system to adaptively control the rotor speed via identification of the mechanical and current-fed system dynamics. Both systems are inherently adaptive as well as self-commissioning. The current controller is a completely general nonlinear controller which can be used together with any drive algorithm. Various advantages of these control schemes over conventional schemes are cited, and the combined speed and current control scheme is compared with the standard vector control scheme

Identification and control of induction machines using artificial neural networks

This paper proposes the use of artificial neural networks (ANNs) to identify and control an induction machine. Two systems are presented: a system to adaptively control the stator currents via identification of the electrical dynamics; and a system to adaptively control the rotor speed via identification of the mechanical and current-fed system dynamics. Various advantages of these control schemes over other conventional schemes are cited and the performance of the combined speed and current control scheme is compared with that of the standard vector control scheme

Alleviating ‘overfitting’ via genetically-regularised neural network

A hybrid genetic algorithm/scaled conjugate gradient regularisation method is designed to alleviate ANN `over-fitting'. In application to day-ahead load forecasting, the proposed algorithm performs better than early-stopping and Bayesian regularisation, showing promising initial results.

Making sense of brain tumour: A qualitative investigation of personal and social processes of adjustment

This study investigated personal and social processes of adjustment at different stages of illness for individuals with brain tumour. A purposive sample of 18 participants with mixed tumour types (9 benign and 9 malignant) and 15 family caregivers was recruited from a neurosurgical practice and a brain tumour support service. In-depth semi-structured interviews focused on participants’ perceptions of their adjustment, including personal appraisals, coping and social support since their brain tumour diagnosis. Interview transcripts were analysed thematically using open, axial and selective coding techniques. The primary theme that emerged from the analysis entailed “key sense making appraisals”, which was closely related to the following secondary themes: (1) Interactions with those in the healthcare system, (2) reactions and support from the personal support network, and (3) a diversity of coping efforts. Adjustment to brain tumour involved a series of appraisals about the illness that were influenced by interactions with those in the healthcare system, reactions and support from people in their support network, and personal coping efforts. Overall, the findings indicate that adjustment to brain tumour is highly individualistic; however, some common personal and social processes are evident in how people make sense of and adapt to the illness over time. A preliminary framework of adjustment based on the present findings and its clinical relevance are discussed. In particular, it is important for health professionals to seek to understand and support individuals’ sense-making processes following diagnosis of brain tumour.

Functional role of cell surface CUB domain-containing protein 1 in tumour cell dissemination

The function of CUB domain-containing protein 1 (CDCP1), a recently described transmembrane protein expressed on the surface of hematopoietic stem cells and normal and malignant cells of different tissue origin, is not well defined. The contribution of CDCP1 to tumor metastasis was analyzed by using HeLa carcinoma cells overexpressing CDCP1 (HeLa-CDCP1) and a high-disseminating variant of prostate carcinoma PC-3 naturally expressing high levels of CDCP1 (PC3-hi/diss). CDCP1 expression rendered HeLa cells more aggressive in experimental metastasis in immunodeficient mice. Metastatic colonization by HeLa-CDCP1 was effectively inhibited with subtractive immunization-generated, CDCP1-specific monoclonal antibody (mAb) 41-2, suggesting that CDCP1 facilitates relatively late stages of the metastatic cascade. In the chick embryo model, time- and dose-dependent inhibition of HeLa-CDCP1 colonization by mAb 41-2 was analyzed quantitatively to determine when and where CDCP1 functions during metastasis. Quantitative PCR and immunohistochemical analyses indicated that CDCP1 facilitated tumor cell survival soon after vascular arrest. Live cell imaging showed that the function-blocking mechanism of mAb 41-2 involved enhancement of tumor cell apoptosis, confirmed by attenuation of mAb 41-2–mediated effects with the caspase inhibitor z-VAD-fmk. Under proapoptotic conditions in vitro, CDCP1 expression conferred HeLa-CDCP1 cells with resistance to doxorubicin-induced apoptosis, whereas ligation of CDCP1 with mAb 41-2 caused additional enhancement of the apoptotic response. The functional role of naturally expressed CDCP1 was shown by mAb 41-2–mediated inhibition of both experimental and spontaneous metastasis of PC3-hi/diss. These findings confirm that CDCP1 functions as an antiapoptotic molecule and indicate that during metastasis CDCP1 facilitates tumor cell survival likely during or soon after extravasation.

Identification of circulating tumour cells in early stage breast cancer patients using multi marker immunobead RT-PCR

Introduction The ability to screen blood of early stage operable breast cancer patients for circulating tumour cells is of potential importance for identifying patients at risk of developing distant relapse. We present the results of a study of the efficacy of the immunobead RT-PCR method in identifying patients with circulating tumour cells. Results Immunomagnetic enrichment of circulating tumour cells followed by RT-PCR (immunobead RT-PCR) with a panel of five epithelial specific markers (ELF3, EPHB4, EGFR, MGB1 and TACSTD1) was used to screen for circulating tumour cells in the peripheral blood of 56 breast cancer patients. Twenty patients were positive for two or more RT-PCR markers, including seven patients who were node negative by conventional techniques. Significant increases in the frequency of marker positivity was seen in lymph node positive patients, in patients with high grade tumours and in patients with lymphovascular invasion. A strong trend towards improved disease free survival was seen for marker negative patients although it did not reach significance (p = 0.08). Conclusion Multi-marker immunobead RT-PCR analysis of peripheral blood is a robust assay that is capable of detecting circulating tumour cells in early stage breast cancer patients.

Long term survival following the detection of circulating tumour cells in head and neck squamous cell carcinoma

Background Techniques for detecting circulating tumor cells in the peripheral blood of patients with head and neck cancers may identify individuals likely to benefit from early systemic treatment. Methods Reconstruction experiments were used to optimise immunomagnetic enrichment and RT-PCR detection of circulating tumor cells using four markers (ELF3, CK19, EGFR and EphB4). This method was then tested in a pilot study using samples from 16 patients with advanced head and neck carcinomas. Results Seven patients were positive for circulating tumour cells both prior to and after surgery, 4 patients were positive prior to but not after surgery, 3 patients were positive after but not prior to surgery and 2 patients were negative. Two patients tested positive for circulating cells but there was no other evidence of tumor spread. Given this patient cohort had mostly advanced disease, as expected the detection of circulating tumour cells was not associated with significant differences in overall or disease free survival. Conclusion For the first time, we show that almost all patients with advanced head and neck cancers have circulating cells at the time of surgery. The clinical application of techniques for detection of spreading disease, such as the immunomagnetic enrichment RT-PCR analysis used in this study, should be explored further.

The sample complexity of pattern classification with neural networks: the size of the weights is more important than the size of the network

Sample complexity results from computational learning theory, when applied to neural network learning for pattern classification problems, suggest that for good generalization performance the number of training examples should grow at least linearly with the number of adjustable parameters in the network. Results in this paper show that if a large neural network is used for a pattern classification problem and the learning algorithm finds a network with small weights that has small squared error on the training patterns, then the generalization performance depends on the size of the weights rather than the number of weights. For example, consider a two-layer feedforward network of sigmoid units, in which the sum of the magnitudes of the weights associated with each unit is bounded by A and the input dimension is n. We show that the misclassification probability is no more than a certain error estimate (that is related to squared error on the training set) plus A3 √((log n)/m) (ignoring log A and log m factors), where m is the number of training patterns. This may explain the generalization performance of neural networks, particularly when the number of training examples is considerably smaller than the number of weights. It also supports heuristics (such as weight decay and early stopping) that attempt to keep the weights small during training. The proof techniques appear to be useful for the analysis of other pattern classifiers: when the input domain is a totally bounded metric space, we use the same approach to give upper bounds on misclassification probability for classifiers with decision boundaries that are far from the training examples.

Neural network learning:theoretical foundations

This important work describes recent theoretical advances in the study of artificial neural networks. It explores probabilistic models of supervised learning problems, and addresses the key statistical and computational questions. Chapters survey research on pattern classification with binary-output networks, including a discussion of the relevance of the Vapnik Chervonenkis dimension, and of estimates of the dimension for several neural network models. In addition, Anthony and Bartlett develop a model of classification by real-output networks, and demonstrate the usefulness of classification with a "large margin." The authors explain the role of scale-sensitive versions of the Vapnik Chervonenkis dimension in large margin classification, and in real prediction. Key chapters also discuss the computational complexity of neural network learning, describing a variety of hardness results, and outlining two efficient, constructive learning algorithms. The book is self-contained and accessible to researchers and graduate students in computer science, engineering, and mathematics