#AQUAMACS IDO FREE#

Our model proposes that human memory uses indirect associations to learn part-of-speech and that the basic associative mechanisms of memory and learning support knowledge of both semantics and grammatical structure. Word associations at four degrees of separation also improve the ability of the model to construct grammatical English sentences. We find that word associations at four degrees of separation increase the similarity assigned by the model to English words that share part-of-speech or syntactic type. By recursively adding higher levels of representations to a computationalвЂљ holographic model of semantic memoryвЂљ we construct a distributional model sensitive to associations between words at arbitrary degrees of separation. Distributional models are sensitive to associations between words at one degree of separationвЂљ such as вЂtigerвЂ™ and вЂstripesвЂ™вЂљ or two degrees of separationвЂљ such as вЂsoarвЂ™ and вЂflyвЂ™. We examine what grammatical information is encoded in distributional models and investigate the role of indirect associations. word embeddings) represent a wordвЂ™s meaning in terms of its relationships with all other words.

HDM provides a flexibleвЂљ scalable alternative to symbolic cognitive architectures at a level of description that bridges symbolicвЂљ quantumвЂљ and neural models of cognition.Д¬omputational models of distributional semantics (a.k.a.

#AQUAMACS IDO FREE#

HDM accounts for primacy and recency effects in free recallвЂљ the fan effect in recognitionвЂљ probability judgementsвЂљ and human performance on an iterated decision task. Our proposed modelвЂљ Holographic Declarative Memory (HDM)вЂљ is a vector-space model based on distributional semantics. We posit that cognitive architecturesвЂљ if implemented in a vector-space modelвЂљ represent a usefulвЂљ explanatory model of the internal representations of otherwise opaque neural architectures. Vector-symbolic architecturesвЂљ where symbols are represented as vectorsвЂљ bridge the gap between the two approaches. Symbolic cognitive architectures can capture the complexities of high-level cognition and provide human-readableвЂљ explainable modelsвЂљ but scale poorly to naturalisticвЂљ non-symbolicвЂљ or big data. HoweverвЂљ while neural networks have an impressive ability to process data to find patternsвЂљ they do not typically model high-level cognitionвЂљ and it is often unclear how they work. High-dimensional vector spaces underlie the success of modern machine learning techniques based on deep learning. We demonstrate that the key components of cognitive architectures-declarative and procedural memory-and their key capabilities-learningвЂљ memory retrievalвЂљ probability judgementвЂљ and utility estimation-can be implemented as algebraic operations on vectors and tensors in a high-dimensional space using a distributional semantics model.

Aquamacs ido