Algorithms and Theory for Multiple-Source Adaptation
Authors: Judy Hoffman, Mehryar Mohri, Ningshan Zhang
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We report the results of a series of experiments with real-world datasets. We find that our algorithm outperforms competing approaches by producing a single robust model that performs well on any target mixture distribution. Altogether, our theory, algorithms, and empirical results provide a full solution for the multiple-source adaptation problem with very practical benefits. |
| Researcher Affiliation | Collaboration | Judy Hoffman CS Department UC Berkeley Berkeley, CA 94720 jhoffman@eecs.berkeley.edu Mehryar Mohri Courant Institute and Google New York, NY 10012 mohri@cims.nyu.edu Ningshan Zhang New York University New York, NY 10012 nzhang@stern.nyu.edu |
| Pseudocode | No | The paper describes the steps of the DC-programming algorithm in prose, but it does not provide a formally labeled 'Pseudocode' block or 'Algorithm' figure. |
| Open Source Code | No | The paper does not provide any statement or link indicating the availability of open-source code for the described methodology. |
| Open Datasets | Yes | We used the sentiment analysis dataset proposed by Blitzer et al. (2007) and used for multiple-source adaptation by Mansour et al. (2008, 2009a). This dataset consists of product review text and rating labels taken from four domains: books (B), dvd (D), electronics (E), and kitchen (K), with 2,000 samples for each domain. ... We considered two real-world domain adaptation tasks: a generalization of a digit recognition task and a standard visual adaptation Office dataset. ... Google Street View House Numbers (SVHN), MNIST, and USPS. ... Our next experiment used the standard visual adaptation Office dataset, which has 3 domains: amazon, webcam, and dslr. The dataset contains 31 recognition categories of objects commonly found in an office environment. |
| Dataset Splits | No | The paper mentions using 'full training sets per domain' and 'a small subset of 200 real image-label pairs from each domain to learn the parameter z', and specific training examples for the Office dataset, but it does not explicitly detail a separate validation split or how it was used in the model training process. |
| Hardware Specification | No | The paper does not specify any particular hardware (e.g., GPU model, CPU type, memory) used for running the experiments. It only mentions that the optimization 'may be solved on a single CPU'. |
| Software Dependencies | No | The paper mentions general software components like 'support vector regression', 'convolutional neural network (CNN)', 'Alex Net', and 'Conv Net' but does not provide specific version numbers for any libraries, frameworks, or solvers used. |
| Experiment Setup | No | The paper mentions training 'base hypotheses using support vector regression with the same hyper-parameters as in (Mansour et al., 2008, 2009a)' but does not explicitly state these hyper-parameters within the paper itself. It also mentions training data sizes but not other configuration details like learning rates, batch sizes, or optimizers. |