| layout | title | permalink |
|---|---|---|
page |
about |
/about/ |
How is memory gradually built up during one’s conscious, or even unconscious, life and thought? My guess is that everything we experience is classified and registered on very many parallel channels in different locations.
Stanislaw Ulam
Pentti Kanerva’s memory model was a revelation for me: it was the very first piece of research I had ever come across that made me feel I could glimpse the distant goal of understanding how the brain works as a whole. It gave me a concrete sense for how familiar mental phenomena could be thought of as distributed patterns of micro-events, thanks to beautiful mathematics.
Douglas Hofstadter
The emphasis might have been on the breadth of topics, in detriment of depth here or there. The work on, say, reinforcement learning, is most definitely not the last word we will see on the subject, but a challenge left for readers to contemplate. But this is due to our research group’s enthusiasm for the topic; we do indeed believe that SDM is — if not correct — extremely close to a full scientific understanding of human long-term memory. If so, it is such a monumental achievement that we want readers to be able to see all of what we see and imagine the vastness of possibilities.
Ralph Waldo Emerson once said: do not go where the path may lead. Go, instead, where there is no path, and leave a trail. Professor Pentti Kanerva has left the trail. It is my job to illuminate it and to pave it and to clear it; to try to deliver an easier pathway for the next generation. Some essays completely shut the door close at the end; this one intends to leave it wide open. As the reader might have noticed, this final section does not read as an analysis of the work done; it reads, instead, as a desideratum, a prologue, a yearning for others to join me in imagining the shape of things to come.
Marcelo S. Brogliato
Exploring the AI potential of Sparse Distributed Memory.
Here you will find all sorts of resources to start (or enhance) your research (or industrial) work on models related to Pentti Kanerva's Sparse Distributed Memory.
Some things are as silly as a small jupyter notebook, toying around with some experiment, or with the mathematics of SDM. Others include a full-blown massively parallel GPGU Cross-platform open-source buzzword-complete framework.
* A massively parallel framework in which users can test cognitive neuroscience theories;
* a full documentation of the framework;
* An open-source book, with a visual tutorial of the system, its relation to neuroscience and psychology, and numerous new scientific results;
* A collection of SDM resources, links, chronology of papers, etc;
* Slides following the Jupyter notebooks used;
* A video-based mini-course on SDM.
This is a framework which can be adapted to any usage of a Sparse Distributed Memory (Kanerva, 1988): new theoretical results, new applications, new machine learning models, etc.
We have been working on Pentii Kanerva's Sparse Distributed Memory:
- The first paper considered the range of dimensions that an SDM should have were it to respect i) chunking-by-averaging, and ii) the "magic number 7";
- a second paper studied the critical distance as the memory becomes saturated;
- a third paper (underway) studies interaction effects between different attractors, and
- a fourth paper will document this highly-palallel SDM framework developed by Dr. Marcelo Brogliato.
We would really like to ask you for feedback, and, should you find it useful, please consider a link or citation:
Brogliato, M.S.; Linhares, A. (2017) Sparse Distributed Memory: a reference implementation. Working Paper, FGV, Vialink.
To generate the library and run some tests:
cd src/
make
make tests
./test1
To run Python tests:
python tests.py
In dev mode, there should be a symbolic link from src/libsdm.so to sdm/_libsdm.so. If it does not exist, create one running:
cd sdm/
ln -s ../src/libsdm.so _libsdm.so
This framework has the following dependencies: libbsd and libOpenCL.