Development of the SP machine
share
This paper describes the main things that need to be done to develop the "SP machine", based on the "SP theory of intelligence" and its realisation in the "SP computer model". The SP machine may be developed initially as a software virtual machine with high levels of parallel processing, hosted on a high-performance computer, or driven by high-parallel search processes in any of the leading search engines. An easy-to-use user interface is needed with facilities for visualisation of knowledge structures and processing. The system needs to be generalised to work with patterns in two dimensions, including the display of multiple alignments with 2D patterns. Research is needed into how the system may discover low-level features in speech and visual images. Existing strengths of the SP system in the processing of natural language may be developed towards the understanding of natural language and its production from meanings. This may be done most effectively in conjunction with the development of existing strengths of the SP system in unsupervised learning. Existing strengths of the SP system in pattern recognition may be developed for computer vision. Further work is needed on the representation of numbers and the performance of arithmetic processes in the SP system. A computer model is needed of SP-neural drawing on the existing conceptual model. When the SP machine is relatively mature, new hardware may be developed to exploit opportunities to increase the efficiency of computations. There is potential for the SP machine to be applied on relatively short timescales in such areas as information storage and retrieval, with intelligence, software engineering with or without automation or semi-automation, and information compression.
READ FULL TEXT