Balancing Appearance and Context in Sketch Interpretation

Authors: Yale Song, Randall Davis, Kaichen Ma, Dana L. Penney

IJCAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We calibrate the utility of different forms of context, describing experiments with Conditional Random Fields trained and tested using a variety of features.
Researcher Affiliation Collaboration Yale Song1, Randall Davis2, Kaichen Ma3, Dana L. Penney4 1Yahoo Research, 2Massachusetts Institute of Technology, 3Google Inc., 4Lahey Hospital and Medical Center
Pseudocode Yes Algorithm 1 Detecting Overwriting and Augmentation; Algorithm 2 Repair Strategy for Under-segmentation; Algorithm 3 Repair Strategy for Over-segmentation
Open Source Code No The paper does not provide any statements about releasing source code or links to a code repository.
Open Datasets No We used a set of 2,024 clock drawings collected from clinical facilities: 1,654 clocks drawn by healthy participants and 370 clocks randomly selected from mildly cognitively impaired participants. We have accumulated a database of several thousand tests from both healthy and cognitively impaired subjects, a unique dataset of real-world drawing behavior.
Dataset Splits Yes To avoid over-fitting, we used an l2 regularization with its scale term set at 0.01, chosen based on 10-fold cross validation.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., CPU, GPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions software components like 'boosted logistic regression classifier' and 'sketched symbol recognizer of [Ouyang and Davis, 2009b]' but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes To avoid over-fitting, we used an l2 regularization with its scale term set at 0.01, chosen based on 10-fold cross validation.