AdaFlock: Adaptive Feature Discovery for Human-in-the-loop Predictive Modeling
Authors: Ryusuke Takahama, Yukino Baba, Nobuyuki Shimizu, Sumio Fujita, Hisashi Kashima
AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The results of our experiments conducted using real datasets indicate that Ada Flock successfully discovers informative features with fewer iterations and achieves high classification accuracy. |
| Researcher Affiliation | Collaboration | Ryusuke Takahama scouty Inc. r.takahama7591@gmail.com Yukino Baba Kyoto University baba@i.kyoto-u.ac.jp Nobuyuki Shimizu, Sumio Fujita Yahoo Japan Corporation {nobushim, sufujita}@yahoo-corp.jp Hisashi Kashima Kyoto University; RIKEN Center for AIP kashima@i.kyoto.ac.jp |
| Pseudocode | Yes | Algorithm 1 Ada Flock |
| Open Source Code | No | No explicit statement or link indicating that the source code for the Ada Flock methodology is open-source or publicly available. |
| Open Datasets | Yes | We obtained 200 paintings by Monet1 and 200 paintings by Sisley2. Each picture was cropped from each side by 30 pixels to remove signatures. The smiles dataset contains videos of spontaneous or posed smiles of enjoyment. We randomly selected 200 spontaneous smile videos and 200 posed smile videos from the Uv A-NEMO Smile Database (Dibeklio glu, Salah, and Gevers 2012).1http://www.claudemonetgallery.org/ 2http://www.alfredsisley.org/ 3http://www.uva-nemo.org |
| Dataset Splits | No | No explicit mention of a validation set or a training/validation/test split. The paper states: 'We used 200 examples for training, from which 100 were positive and the others were negative; the remaining 200 examples were used for testing.' |
| Hardware Specification | No | No specific hardware details (like CPU/GPU models, memory) are provided for running the experiments. |
| Software Dependencies | No | The paper mentions software like 'Ada Boost classifier', 'decision tree', and 'Inception-v3 model' but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | Yes | We fix the number of iterations to a rather large value (i.e., T = 10) to investigate the change of the accuracies according to the number of iterations. We set the number of selected examples at each iteration to m = 20 and the number of feature definitions at each iteration to k = 10 because we consider these settings are convenient for workers. |