Allison Petrosino

Papers

Total Citations

H-Index

About

Allison Petrosino’s research lies at the intersection of machine learning, cognitive science, and robotics, with a focus on how artificial systems can learn language and concepts more like humans do. Her most-cited work, *“Toward Fast Mapping for Robot Adjective Learning,”* explores how robots can acquire adjective meanings from minimal exposure—mimicking the human phenomenon of “fast mapping,” where children learn new words after just a few contrasts with known terms. This preliminary study demonstrated that machine learners could leverage such contrasts in unconstrained speech, opening a path toward more intuitive human-robot interaction. In parallel, Petrosino’s paper *“Using information gain to build meaningful decision forests for multilabel classification”* introduced “Gain-Based Separation,” a novel heuristic that adapts standard decision tree algorithms to handle objects with multiple simultaneous labels. This work addresses a fundamental challenge in multilabel classification, where information gain can be misleading when a single class dominates. Though her citation counts are modest (7 each), these contributions are notable for their interdisciplinary ambition—bridging developmental psychology and machine learning—and for laying early groundwork in efficient, human-like learning for autonomous systems.