learning algorithms

A novel method for sampling a class of subsequential string transductions encoding homomorphisms allows rigorous testing of learning models’ capacity for compositionality.

This chapter examines the brief but vibrant history of learnability in phonology.

A book chapter on mathematical theories of language and learning, and their consequences for linguistic cognition studies.

We derive the well-studied subregular classes of formal languages, which computationally characterize natural language typology, purely from the perspective of algorithmic learning problems.

We comment on non-human animals’ ability to learn syntactic vs phonological dependencies in pattern-learning experiments.

We comment on mathematical fallacies present in artificial grammar learning experiments and suggest how to integrate psycholinguistic and mathematical results.

We analyze the expressivity of a variety of recurrent encoder-decoder networks, showing they are limited to learning subsequential functions, and connecting RNNs with attention mechanisms to a class of deterministic 2-way transducers.

I provide a vector space characterization of the Star-Free and Locally Threshold testable classes of formal languages, over arbitrary data structures.

We describe a partial order on the space of model-theoretic constraints and a learning algorithm for constraint inference.

We caution about confusing ignorance of biases with absence of biases in machine learning and linguistics, especially for neural networks.