Domain Adaptation under Open Set Label Shift

Authors: Saurabh Garg, Sivaraman Balakrishnan, Zachary Lipton

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments across numerous semi-synthetic benchmarks on vision, language, and medical datasets demonstrate that our methods consistently outperform open set domain adaptation baselines, achieving 10 25% improvements in target domain accuracy.
Researcher Affiliation Academia Saurabh Garg, Sivaraman Balakrishnan, Zachary C. Lipton Machine Learning Department, Department of Statistics and Data Science, Carnegie Mellon University {sgarg2,sbalakri,zlipton}@andrew.cmu.edu
Pseudocode Yes Algorithm 1 Positive and Unlabeled learning post Label Shift Estimation (PULSE) framework
Open Source Code Yes Code is available at https://github.com/acmi-lab/Open-Set-Label-Shift
Open Datasets Yes For vision, we use CIFAR10, CIFAR100 [40] and Entity30 [61]. For language, we experiment with Newsgroups-20 (http://qwone.com/~jason/20Newsgroups/) dataset. Additionally, inspired by applications of OSLS in biology and medicine, we experiment with Tabula Muris [17] (Gene Ontology prediction), Dermnet (skin disease prediction https://dermnetnz.org/), and Break His [66] (tumor cell classification).
Dataset Splits Yes We repeat the same process on iid hold out data to obtain validation data with no target labels.
Hardware Specification Yes All experiments are run on an internal compute cluster with NVIDIA V100 GPUs.
Software Dependencies No The paper mentions using PyTorch for implementation but does not specify its version or any other software dependencies with version numbers (e.g., 'PyTorch (Paszke et al., 2019)' is a citation, not a version number).
Experiment Setup Yes We use default hyperparameters for all methods. For OSDA methods, we use default method specific hyperparameters introduced in their works. We provide precise details about hyperparameters, OSLS setup for each dataset and code in App. F.3.