Abstract :
There has been a lot of interest in adaptive symbolic and neural agents for different tasks, for instance speech/language integration and image/text integration in various multimedia applications. Hybrid neural symbolic methods have been shown to be able to reach a level where they can actually be further developed in real-world scenarios. A combination of symbolic and neural agents is possible in various neural symbolic processing architectures, which contain both symbolic and neural agents appropriate for to a specific task, e.g. integrating speech, text and images for multimedia. We concentrate on general principles of neural and hybrid architectures for multimedia in general. From the perspective of knowledge engineering, hybrid symbolic/neural agents are advantageous since different mutually complementary properties can be combined. Symbolic representations have advantages with respect to easy interpretation, explicit control, fast initial coding, dynamic variable binding and knowledge abstraction. On the other hand, neural agents show advantages for gradual analog plausibility, learning, robust fault-tolerant processing, and generalization to similar input. Since these advantages are mutually complementary, a hybrid symbolic neural architecture can be useful if different processing strategies have to be supported
