Multi-Source Domain Adaptation for Text Classification via DistanceNet-Bandits

Authors: Han Guo, Ramakanth Pasunuru, Mohit Bansal7830-7838

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct experiments on popular sentiment analysis datasets with several diverse domains and show that our Distance Net model, as well as its dynamic bandit variant, can outperform competitive baselines in the context of unsupervised domain adaptation.
Researcher Affiliation Academia Han Guo, Ramakanth Pasunuru, Mohit Bansal UNC Chapel Hill {hanguo, ram, mbansal}@cs.unc.edu
Pseudocode No The paper describes the models and methods using mathematical formulas and descriptive text, but does not include any pseudocode or algorithm blocks.
Open Source Code No The paper does not provide an explicit statement or a link to open-source code for the described methodology.
Open Datasets Yes Dataset: We evaluate our methods using the datasets collected by Liu, Qiu, and Huang (2017), which contains 16 datasets of product reviews (Blitzer, Dredze, and Pereira 2007) and movie reviews (Maas et al. 2011; Pang and Lee 2005), where the task is to classify these reviews as positive or negative.
Dataset Splits No Following Ruder and Plank (2017) and Bousmalis et al. (2016), we chose to use a small number of target domain examples as validation set (for both tuning as well as providing rewards for the multi-armed bandit controller). While a validation set is used, the paper only states 'a small number of target domain examples' without providing specific percentages, sample counts, or a detailed splitting methodology.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU models, CPU specifications, memory) used to run the experiments.
Software Dependencies No The paper mentions several software components and libraries (e.g., Adam optimizer, Ax, JAX library) but does not provide specific version numbers for any of them, which is required for a reproducible description of ancillary software.
Experiment Setup Yes Our baseline model is similar to that of Liu, Qiu, and Huang (2017). We use a single-layer Bidirectional LSTM-RNN as sentence encoder and a two-layer fully-connected with ReLU non-linearity layer to produce the final model outputs. The word embeddings are initialized with GloVe (Pennington, Socher, and Manning 2014). We train the model using Adam optimizer (Kingma and Ba 2014)... We use the adaptive experimentation platform Ax to tune the rest of the hyperparameters and the search space for these hyperparameters are: learning rate (10^-4, 10^-3), dropout rate (0.25, 0.75), β (0.01, 1.0), and αk (0.0, 1.0).