Since inductive reasoning is the process of inferring a cause and making predictions based on previous observations (right?), wouldn’t the standard HTM sequence memory already be considered an inductive reasoning system?
In your example, the HTM learns the temporal sequence pattern 1, 2, 3, 4, 5. If it sees the sequence 1, 2, 3, 4, X, it uses an inductive inference algorithm (the learned sequence memory of previously observed patterns) to infer the cause and predict the “5”. If it sees a temporal sequence that is different, but semantically similar to, the one it has already learned, then it should also be able to predict something similar to a “5”. That doesn’t exactly address the example you described, but it might essentially answer your general question. However, I’m kind of thinking (might be wrong) that your example is actually describing the process of deduction. After seeing 1, 2, 3, 4, 5, the HTM needs to deduce a hypothesis that these sequences are certain “classes” and are composed of patterns which semantically differ from each other by “1”, whatever “1” is in terms of SDRs representing mathematical constructs. It can then test this hypothesis on the new sequence 6, 7 ,8 ,9, X. Maybe someone else can either correct if I am wrong, or help explain the reasoning idea better. From: nupic [mailto:[email protected]] On Behalf Of Alexandre Vivmond Sent: Tuesday, April 26, 2016 11:06 AM To: Dillon Bender <[email protected]> Subject: HTM/Nupic's ability for inductive reasoning Does HTM's have the ability to do inductive reasoning based on previously seen similar patterns? Here's are my thoughts: say that an input file will only ever include values between 1 and 10. The values 1 and 2 should have a lot of semantic overlap compared to 1 and 10. Then, let HTM learn the pattern 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, ... may hundred times, and at one point suddenly see the pattern 6, 7, 8, 9, X, can HTM inductively reason that the next value X should be 10 given that the pattern is similar to 1, 2, 3, 4, 5 and that the semantic overlap between 4 and 5 is similar to 9 and 10? I tried this exact problem in Nupic and it unfortunately did not manage to predict the never before seen values 6, 7, 8, 9, 10. If HTM/nupic can inductively reason could you provide some examples that should/might work? The reason I'm asking is because I'm curious since the Fox eats example (from the nupic hackathon 2013) is similar to my mentioned problem (in my understanding at least), it managed to correctly predict what the fox eats, even though HTM had never previously learned what foxes eat nor been put in any context other than the SDR's semantic understanding that foxes are similar to other animals that eat rodents.
