Related Experiment Videos
Hinfinity-learning of layered neural networks
1Department of Computer and Information Science, Faculty of Engineering, Iwate University, Morioka 020-8551, Japan. nisiyama@cis.iwate-u.ac.jp
IEEE Transactions on Neural Networks
|February 6, 2008
Summary
This study introduces H(infinity)-learning, a novel algorithm that improves neural network training speed and robustness compared to backpropagation (BP) and extended Kalman filter (EKF) methods.
Area of Science:
- Artificial Intelligence
- Machine Learning
- Neural Networks
Background:
- Backpropagation (BP) learning for neural networks has slow convergence.
- Extended Kalman Filter (EKF) offers faster learning but is sensitive to noise and initial values.
- Robustness issues with EKF necessitate improved learning algorithms.
Purpose of the Study:
- To develop a novel learning rule, H(infinity)-learning, for enhanced neural network training.
- To derive globally and locally optimized learning algorithms based on H(infinity)-learning.
- To improve robustness against noise and initial value variations in neural network learning.
Main Methods:
- Proposed H(infinity)-learning as a new learning rule.
- Derived globally and locally optimized algorithms from H(infinity)-learning.
- Analyzed learning behavior using computer simulations.
Main Results:
- H(infinity)-learning algorithms demonstrate improved learning behavior.
- The derived algorithms show enhanced robustness to initial values and noise.
- Comparative analysis indicates competitive performance and computational cost against BP and EKF.
Conclusions:
- H(infinity)-learning offers a promising alternative for faster and more robust neural network training.
- The derived algorithms provide optimized solutions for practical applications.
- Further research can explore broader applications of H(infinity)-learning.
Related Concept Videos
01:18Introduction to Learning
Learning is the process of acquiring knowledge or skills through practice or experience, leading to long-lasting behavioral changes. This acquisition occurs through interaction with the environment and requires practice or experience. For instance, mastering a skill such as surfing requires considerable practice and experience, highlighting the essential role of repeated interactions with the environment in learning.
In contrast to learned behaviors, unlearned behaviors such as crying, sexual...
In contrast to learned behaviors, unlearned behaviors such as crying, sexual...
01:27Associative Learning
Associative learning is a fundamental concept in behavioral psychology, wherein a connection is established between two stimuli or events, leading to a learned response. This process is critical in understanding how behaviors are acquired and modified. Conditioning, the mechanism through which associations are formed, can be divided into two main types: classical conditioning and operant conditioning, each elucidating different aspects of associative learning.
Classical conditioning, also known...
Classical conditioning, also known...
01:12Observational Learning
Albert Bandura's observational learning, also known as imitation or modeling, occurs when a person observes and imitates another's behavior. It is a quicker process than operant conditioning. A well-known example is the Bobo doll study, where children who saw an adult acting aggressively towards the doll were more likely to act aggressively when left alone, compared to those who observed a nonaggressive adult. Many psychologists view observational learning as a form of latent learning because...
01:24Convolution: Math, Graphics, and Discrete Signals
In any LTI (Linear Time-Invariant) system, the convolution of two signals is denoted using a convolution operator, assuming all initial conditions are zero. The convolution integral can be divided into two parts: the zero-input or natural response and the zero-state or forced response, with t0 indicating the initial time.
To simplify the convolution integral, it is assumed that both the input signal and impulse response are zero for negative time values. The graphical convolution process...
To simplify the convolution integral, it is assumed that both the input signal and impulse response are zero for negative time values. The graphical convolution process...