This paper aims to take general tensors as inputs for supervised learning. A supervised tensor learning (STL) framework is established for convex optimization based learning techniques such as support vector machines ...
详细信息
This paper aims to take general tensors as inputs for supervised learning. A supervised tensor learning (STL) framework is established for convex optimization based learning techniques such as support vector machines (SVM) and minimax probability machines (MPM). Within the STL framework, many conventional learning machines can be generalized to take n/sup th/-order tensors as inputs. We also study the applications of tensors to learning machine design and feature extraction by linear discriminant analysis (LDA). Our method for tensor based feature extraction is named the tenor rank-one discriminant analysis (TR1DA). These generalized algorithms have several advantages: 1) reduce the curse of dimension problem in machine learning and data mining; 2) avoid the failure to converge; and 3) achieve better separation between the different categories of samples. As an example, we generalize MPM to its STL version, which is named the tensor MPM (TMPM). TMPM learns a series of tensor projections iteratively. It is then evaluated against the original MPM. Our experiments on a binary classification problem show that TMPM significantly outperforms the original MPM.
The key measurement instruments in the control of the base metal flotation are the on-line elemental analyzers like the Courier X-ray fluorescence analyzer. The calibration of the analyser is one of the key factors in...
详细信息
The key measurement instruments in the control of the base metal flotation are the on-line elemental analyzers like the Courier X-ray fluorescence analyzer. The calibration of the analyser is one of the key factors in the operation of the X-ray fluorescence analyser. The quality of the analyses is limited by the quality of the calibration of the assay calculation model. The co-linearity of the measurement information is problematic for the multi-linear regression used in the assay calculation and calibration. In this study two new regression methods are introduced for the assay calculation and calibration: Principal Component Regression and Partial Least Squares regression. These methods utilize the idea of internal structure of data - latent structure that handles the co-linearity internally.
With the increasing use of sensors and actuators in technical systems and knowledge-intensive services the need for processing the information captured by these sensors and "making sense" out of it increases...
详细信息
With the increasing use of sensors and actuators in technical systems and knowledge-intensive services the need for processing the information captured by these sensors and "making sense" out of it increases. Knowledge fusion is supposed to contribute to this field since it aims at integrating knowledge from different sources. Development of knowledge fusion solutions is a complex task which can be compared to systems and software development. As in other development areas there is a need for efficient development processes which can be supported by reusing solution parts, such as patterns or components. The paper brings together experiences from knowledge fusion subsystem development and from design of knowledge fusion patterns. The main contributions of this paper are (1) a real-world application scenario presenting typical requirements to knowledge fusion systems, (2) application of knowledge fusion patterns from context-based decision support to situation recognition, (3) recommendations from this application case.
A control systems engineering approach, employing a two-level overall system architecture and different but compatible formalisms for system representation on the upper and lower levels, has been investigated in detai...
详细信息
This paper analyzes the outage probability of opportunistic decode-and-forward relaying network over asymmetric fading channels where cooperative nodes are located at distributed locations and signals experience asymm...
详细信息
Deep neural networks (DNNs) are widely used in fields like computer vision and natural language processing. A key component of DNN training is the optimizer. SGD-Momentum is popular in many DNN methodologies, such as ...
详细信息
Fault diagnosis and condition monitoring can be based on analytical process models, explicit knowledge or direct process measurements. When the structure of the process is unknown, the data-based approach based on the...
详细信息
Fault diagnosis and condition monitoring can be based on analytical process models, explicit knowledge or direct process measurements. When the structure of the process is unknown, the data-based approach based on the measurements is the most practical one. This is the case also with X-ray fluorescence analyzers. Appropriate analyzer calibration is a key factor in obtaining the best performance from the analysis system. As the mineralogy of the processed material is slowly changing the calibration must be maintained over time. The possible indicators for the calibration validity are studied in this paper.
Handwriting recognition is one of the most important tasks in computer vision. Despite the variety of methods and algorithms used for text recognition in various languages, the recognition of handwritten Russian words...
详细信息
This paper proposes a new adaptive-gain recurrent neural network (AG-RNN) to effectively cope with the joint-angle drift issues in redundant manipulators. Specifically, a joint-angle drift-free with the feedback contr...
详细信息
This paper presents a new approach for approximate identification under the H 2 criterion where the model is parametrized in a basis generated by apriori knowledge on the location of system poles. A frequency domain i...
详细信息
This paper presents a new approach for approximate identification under the H 2 criterion where the model is parametrized in a basis generated by apriori knowledge on the location of system poles. A frequency domain identification procedure is proposed and it is applied to obtain models for the MIT interferometer testbed.
暂无评论