My interests in M4L are on the development of the network and the research that will result. With regard to the network, I am very enthusiastic about the creation of an interdisciplinary research community in this space, and as a result I am particularly looking forward to the success of our Summer school, or schools, in inspiring the next generation of scientists at the formative stage of their PhD study. In the same vein, I would like to see, in a prominent forum, a 'white paper' statement of the aims of M4L, hopefully conveying the excitement of this particular confluence of disciplines. From this point of view, the benefits of M4L should be visible for many years to come, and we should note that the success criteria of the network should stretch well beyond its two-year lifespan.
In terms of research topics, several are intriguing. First there is the nature of memory, indexing and retrieval. I have always been interested in the notion of creating computer systems that are capable of associative linking. Associative memory is a very efficient, content-based method of retrieving information, and of course characteristic of human memory. There have been great strides in the associative retrieval of artificial memory in the last few years, but this is one area where we might justifiably get excited about the potential input to computer science from the neurosciences.
Multimedia representations of knowledge have always been of great interest to me, and of course they pose particular problems in computing. For example, ##how do we know that a series of multimedia memories of some event, say, across multiple modalities, are actually all about that same event? And how do we perform such inferences in real time at retrieval time?## Again, neuroscientific understanding of human memory could provide invaluable input.
In each of these cases, interdisciplinary study looks to be vital, to understand the neuroscientific and psychological characteristics of human memory, to relate these to more logical features of the abstract indexing problems set by multimedia representations and content-based retrieval, and then to use the enhanced understanding of the abstract indexing issue to improve our computational techniques.
Finally, much M4L research demands good databases and corpuses. The M4L network is likely to be involved in the production of a corpus of some form, and the value of that effort, though somewhat less glamorous, will also be very high in the context of creating and fostering an interdisciplinary research community.
In terms of research topics, several are intriguing. First there is the nature of memory, indexing and retrieval. I have always been interested in the notion of creating computer systems that are capable of associative linking. Associative memory is a very efficient, content-based method of retrieving information, and of course characteristic of human memory. There have been great strides in the associative retrieval of artificial memory in the last few years, but this is one area where we might justifiably get excited about the potential input to computer science from the neurosciences.
Multimedia representations of knowledge have always been of great interest to me, and of course they pose particular problems in computing. For example, ##how do we know that a series of multimedia memories of some event, say, across multiple modalities, are actually all about that same event? And how do we perform such inferences in real time at retrieval time?## Again, neuroscientific understanding of human memory could provide invaluable input.
In each of these cases, interdisciplinary study looks to be vital, to understand the neuroscientific and psychological characteristics of human memory, to relate these to more logical features of the abstract indexing problems set by multimedia representations and content-based retrieval, and then to use the enhanced understanding of the abstract indexing issue to improve our computational techniques.
Finally, much M4L research demands good databases and corpuses. The M4L network is likely to be involved in the production of a corpus of some form, and the value of that effort, though somewhat less glamorous, will also be very high in the context of creating and fostering an interdisciplinary research community.