Hot answers tagged

4

SPA is used (among other things) for combining (binding) and extracting (unbinding) knowledge representations for processing. This is a (purposely) lossy compression. In the "Learning Rule Generation for Induction" case, the clean-up memory is used to convert a general transformation that is being learned (lots of different transformations convolved together)...


4

Both ACT-R and Spaun are modular, and could be extended to include capabilities of each other. Comparing the functionalities of the two architectures by a simple checklist is not the most appropriate way to compare them. Here are some points to consider: 1) ACT-R is primarily symbolic, whereas Spaun is a neural network. ACT-R is actually making a ...


3

Long-term memory storage and forgetting are the two main features that are found in ACT-R, but are missing from Spaun. Currently, Spaun only has a working memory and a fixed long-term memory; it lacks the ability to store new items in a long-term memory. This is currently being tackled by trying to create better hippocampal models and encoder learning rules....


2

Deep Learning (which is now supposed to be called Differentiable Programming?) is a means of doing inference over a set of data. SPA is a Vector Symbolic Architecture (VSA) implemented using Holographic Reduced Representations (HRR). That is to say, sparse vectors represent symbols and there is a set of operations that can be used to combine them. ...


2

From Wikipedia: Content-addressable memory (CAM) is a special type of computer memory used in certain very-high-speed searching applications. It is also known as associative memory, associative storage, or associative array [...] It compares input search data (tag) against a table of stored data, and returns the address of matching data (or in the ...


2

To begin to answer this question, we must first unpack the concepts in their current context. The NEF makes no prediction about how error is propagated in the brain. It explains how to do computation using vectors in spiking neural networks. Also, it defines how error signals can be used to change how the signal is encoded (take in) and decoded (sent out) ...


2

Choo's OSE model paper does not present evidence that would allow us to say whether or not the model exhibits proactive interference or set size effects. Presumably, this is because it is specifically a model of primacy and recency effects, not a general model of working memory. The thesis scope is given on page two: Two major behavioural effects are ...


2

I do not know for sure, but I believe that the vectors would be created in a domain-specific manner. Vectors in the visual system would be created in a way that is particular to the needs of visual processing, etc. The reason why the vectors are represented as random in the semantic pointer architecture, is because from the point of view of the ...


1

Counting task numbers in Spaun becomes somewhat meaningless due to it's instruction following capabilities. However, the missing task was the stimulus response task, wherein given an image from ImageNet, classify it according to it's given identifier.


1

As explained in the NEF book, all non-linear dynamical systems can be represented in the NEF. The SPA is simply applying semantic meaning and manipulations to the vectors represented by the NEF. The two most prominent examples of this are the motor system and the inference system in Spaun, which are shown in the aforementioned "rapid variable creation" task....


Only top voted, non community-wiki answers of a minimum length are eligible