Learning with Fredholm Kernels
Authors: Qichao Que, Mikhail Belkin, Yusu Wang
NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We proceed to discuss the noise assumption for semi-supervised learning and provide both theoretical and experimental evidence that Fredholm kernels can effectively utilize unlabeled data under the noise assumption. We demonstrate that methods based on Fredholm learning show very competitive performance in the standard semi-supervised learning setting. |
| Researcher Affiliation | Academia | Qichao Que Mikhail Belkin Yusu Wang Department of Computer Science and Engineering The Ohio State University Columbus, OH 43210 {que,mbelkin,yusu}@cse.ohio-state.edu |
| Pseudocode | No | The paper provides mathematical formulations and equations but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain any statement about releasing source code or provide links to a code repository. |
| Open Datasets | Yes | We use the following data (represented by TF-IDF features): (1) 20 news group: it has 11269 documents with 20 classes, and we select the first 10 categories for our experiment. (2) Webkb: the original data set contains 7746 documents with 7 unbalanced classes, and we pick the two largest classes with 1511 and 1079 instances respectively. (3) IMDB movie review: it has 1000 positive reviews and 1000 negative reviews of movie on IMDB.com. (4) Twitter sentiment data from Sem-Eval 2013: it contains 5173 tweets, with positive, neural and negative sentiment. We combine neutral and negative classes to set up a binary classification problem. ... The experiment use subsets of two handwriting digits data sets MNIST and USPS: (1) the one from MNIST contains 10k digits in total with balanced examples for each class, and the one for USPS is the original testing set containing about 2k images. |
| Dataset Splits | Yes | To choose the optimal parameters for each method, we pick the parameters based on their performance on the validation set, while the final classification error is computed on the held-out testing data set. |
| Hardware Specification | No | The paper does not specify any hardware details such as CPU, GPU models, or memory used for experiments. |
| Software Dependencies | No | The paper does not mention specific software versions (e.g., programming languages, libraries, or frameworks with their version numbers) used for the experiments. |
| Experiment Setup | No | The paper states that parameters were chosen based on validation set performance but does not provide concrete hyperparameter values, training configurations, or system-level settings in the main text. |