Proceedings of the 12th IEEE Workshop on Neural Networks for Signal Processing
Cơ quản chủ quản: N/A
Các bài báo tiêu biểu
Signal reconstruction from sampled data using neural network
- Trang 707-715
For reconstructing a signal from sampling data, the method based on Shannon's
sampling theorem is usually employed. The reconstruction error appears when the
signal does not satisfy the Nyquist condition. This paper proposes a new
reconstruction method by using a linear perceptron and multilayer perceptron as
FIR filter. The perceptron, which has weights obtained by learning when adapting
the orig...
#Signal reconstruction #Neural networks #Image reconstruction #Sampling methods #Finite impulse response filter #Frequency #Adaptive filters #Image sampling #Information retrieval #Physics
Ký ức liên kết tạm thời và sự xấp xỉ hàm với bản đồ tự tổ chức Dịch bởi AI
- Trang 109-118
Chúng tôi đề xuất một kỹ thuật lập mô hình nơ-ron không giám sát, gọi là ký ức
liên kết tạm thời dạng vector (VQTAM), cho phép bản đồ tự tổ chức của Kohonen
(SOM) xấp xỉ các ánh xạ động lực phi tuyến trên toàn cầu. Phân tích lý thuyết về
phương pháp VQTAM cho thấy rằng độ sai lệch xấp xỉ giảm đi khi quá trình đào tạo
SOM diễn ra. SOM được so sánh với mạng MLP tiêu chuẩn và mạng RBF trong việc xác
...
#Ký ức liên kết #Xấp xỉ hàm #Mô hình hóa hệ thống sinh học #Thiết bị truyền động thủy lực #Hệ thống động lực phi tuyến #Mô hình toán học #Mô hình dự đoán #Hệ thống điều khiển phi tuyến #Mạng hàm cơ sở bán kính #Roentgenium
An efficient SMO-like algorithm for multiclass SVM
- Trang 297-306
Starting from a reformulation of Cramer and Singer (see Journal of Machine
Learning Research, vol.2, p.265-92, Dec. 2001) multiclass kernel machine, we
propose a sequential minimal optimization (SMO) like algorithm for incremental
and fast optimization of the Lagrangian. The proposed formulation allowed us to
define very effective new pattern selection strategies which lead to better
empirical res...
#Support vector machines #Kernel #Support vector machine classification #Lagrangian functions #Prototypes
A hierarchical feedforward adaptive filter for system identification
- Trang 269-278
An architecture for adaptive filtering based upon the previously introduced
hierarchical least mean square algorithm is proposed. This pyramidal
architecture incorporates sparse connections between the architectural layers
with a certain variable degree of overlapping between the neighboring subfilters
of the same level. A learning algorithm for this class of structures is derived,
based on the ba...
#Adaptive filters #System identification #Signal processing algorithms #Neurons #Finite impulse response filter #Neural networks #Adaptive signal processing #Biomedical signal processing #Least squares approximation #Feedforward neural networks
Local minima effects on the transient performance of non-linear blind equalizers
- Trang 717-726
The computational requirements and the transient performance of several
non-linear blind equalizers are compared in the case of transmission over linear
and non-linear channels. The multilayer perceptron (MLP), the
radial-basis-function network (RBF), the polynomial perceptron (PP) and two
recently proposed non-linear structures (see Destro Filho, J.B., et al., Proc.
GLOBECOM'96, p.196-200, 1996; ...
#Blind equalizers #Hydrogen #Multilayer perceptrons #Filters #Polynomials #Satellite communication #Neural networks #Steady-state #Quadratic programming #Computational modeling
Metric-based model selection for time-series forecasting
- Trang 13-22
Metric-based methods, which use unlabeled data to detect gross differences in
behavior away from the training points, have recently been introduced for model
selection, often yielding very significant improvements over alternatives
(including cross-validation). We introduce extensions that take advantage of the
particular case of time-series data in which the task involves prediction with a
horizo...
#Predictive models #Linear regression #Input variables #Testing #Training data #Machine learning
Neural network-based segmentation of textures using Gabor features
- Trang 365-374
The effectiveness of Gabor filters for texture segmentation is well known. In
this paper, we propose a texture identification scheme, based on a neural
network (NN) using Gabor features. The features are derived from both the Gabor
cosine and sine filters. Through experiments, we demonstrate the effectiveness
of a NN based classifier using Gabor features for identifying textures in a
controlled en...
#Neural networks #Gabor filters #Frequency #Multilayer perceptrons #Multi-layer neural network #Image segmentation #Classification algorithms #Clustering algorithms #Robustness #Satellites
Detection of unusual human behavior in intelligent house
- Trang 697-706
This paper describes a model, based on a Markov process model, of daily human
behavior in an intelligent house where human behavior is observed with small
motion detectors. The number of sensor states is reduced to a few dozen by a
vector quantization method, and transitions within this reduced set of states
are observed. Then, the state transition probability and the transition duration
time dist...
#Humans #Intelligent sensors #Home appliances #Vector quantization #Infrared detectors #Motion detection #Programming #Computer industry #Research and development #Cities and towns
Linear input network for neural network automata model adaptation
- Trang 617-626
The paper describes an experimental investigation of the applicability of linear
input networks (LIN) as a channel and noise adaptation technique for an
application of the Loquendo neural network based speech recognizer in a car
environment. The considered application is an automated call center that
provides traffic information through a voice dialogue system. The connection to
the call center is...
#Neural networks #Automata #Adaptation model #Noise reduction #Working environment noise #Acoustic noise #Speech recognition #Speech enhancement #Automatic speech recognition #Telecommunication traffic
Minimax strategies for training classifiers under unknown priors
- Trang 249-258
Most supervised learning algorithms are based on the assumption that the
training data set reflects the underlying statistical model of the real data.
However, this stationarity assumption is not always satisfied in practice: quite
frequently, class prior probabilities are not in accordance with the class
proportions in the training data set. The minimax approach is based on selecting
the classifi...
#Minimax techniques #Neural networks #Training data #Error analysis #Robustness #Supervised learning #Error probability #Cost function #Proposals