VLSI HARDWARE FOR EXAMPLE-BASED LEARNING

Example-based learning, as performed by neural networks and other appr oximation and classification techniques, is both computationally inten sive and I/O intensive, typically involving the optimization of hundre ds or thousands of parameters during repeated network evaluations over a database of example vectors, Although there is currently no dominan t approach or technique among the various neural networks and learning algorithms, the basic functionality of most neural networks can be co nceptually realized as a multidimensional look-up table, While multidi mensional look-up tables are clearly impractical due to the exponentia l memory requirements, we are pursuing an approach using interpolation based only on the sparse data provided by an initial example database , In particular, we have designed prototype VLSI components for search ing multidimensional example databases for the k closest examples to a n input query as determined by a programmable metric using a massively parallel search, This nearest-neighbor approach can be used directly for classification, or in conjunction with any number of neural networ k algorithms that exploit local fitting, The hardware removes the I/O bottleneck from the learning task by supplying a reduced set of exampl es for localized training or classification, Though nearest-neighbor r etrieval algorithms have efficient software implementations for low-di mensional databases, exhaustive searching is the only effective approa ch for handling high-dimensional data, The parallel VLSI hardware we h ave designed can accelerate the exhaustive search by three orders of m agnitude, We believe this special purpose VLSI will have direct applic ation in systems requiring learning functionality and in accelerating learning applications on large, high-dimensional databases.