1. Hello everyone, welcome to NCRA 7th workshop 2016.
Im Nam Le, PhD
student at NCRA from August 2016. Today Im going to talk about my research topic, namely: Social Genetic Programming: The idea from social learning. It is the first main research theme in my PhD studies 2. I earned both my bachelor and my master in Vietnam. 3. When first came to this direction, my first glance is that Social Learning is huge with a lot of abstract and confusing definitions. Actually, I misunderstood that this project is about combination of genetic programming with PSO, but its not. PSO is just a simple form of learning in social context. Then I met social learning theory by Bandura, and pay attention on that theory. So today I will talk shortly about some definitions and issues in STL. And then I present my first attempt using STL with Genetic Programming in form of grammatical evolution, as well as some further improvement I plan to do in near future. 4. What is STL? First discussed by Alberta Bandura, an American Psychologist in 1974. He believed that social element can result in the development of new learning among individuals, and People can learn new things and develop new behaviors by observing other people. Social learning theory integrated behavioral and cognitive theories of learning in order to provide a comprehensive model that could account for the wide range of learning experiences that occur in the real world. Involve observation, extraction of information from those observations, then making decisions. Reinforcement plays a role in learning but is not entirely responsible for learning. So we know that the main focus in STL by bandura is observational learning. Lets move on next slide. 5. The observational learning starts with an attention process. In a social group, members with interesting qualities are likely to receive more attention than the others. At the same time, informative function determines which characteristics of the models capture attention and which will be ignored .Then, people remember the details of their exemplary behavior with a retention process and practice to reproduce the behavior with a reproduction process. However, even if the attention- retention- reproduction process is finished, the person will not engage in the behavior without motivation. In the motivation process, the learnt behavior, which previously remained unexpressed, will take action when incentives are provided . 6. - Attention: Observers cannot learn unless they pay attention to what's happening around them. This process is influenced by characteristics of the model, such as how much one likes or identifies with the model, and by characteristics of the observer, such as the observer's expectations or level of emotional arousal. - Retention/Memory: Observers must not only recognize the observed behavior but also remember it at some later time. This process depends on the observer's ability to code or structure the information in an easily remembered form or to mentally or physically rehearse the model's actions.- Imitation/Reproduction: Observers must be physically and/intellectually capable of producing the act. In many cases the observer possesses the necessary responses. But sometimes, reproducing the model's actions may involve skills the observer has not yet acquired. It is one thing to carefully watch a circus juggler, but it is quite another to go home and repeat those acts. - Motivation: Coaches also give pep talks, recognizing the importance of motivational processes to learning. 7. Reinforcement plays an important role in observational learning for it distinguishes learning from simply imitating the others. In social learning theory, behavior is regulated by external reinforcement, vicarious reinforcement, and self- reinforcement, among which vicarious reinforcement has a crucial role [1]. Vicarious reinforcement is defined as the adaptation in the behavior of observers when they notice the response consequences of the models. Generally, it includes vicarious positive reinforcement (that observers display an increase in the behavior when they see models get positive consequences), and vicarious punishment (that negative consequences prevent observers from behaving similarly like the models). By the effect of vicarious reinforcement, observers may perform even better than the models. Moreover, observational learning is the primary source of innovation in a social group owing to the following reason. Observers will neither concentrate on a single model, nor absorb all characteristics of the preferred model, but they abstract common features of diverse models to form a behavior rule or combine different attributes of the models to develop distinct personalities. The more diverse the models are, the more likely the observers exhibit creative, innovative behavioral patterns.