Self-Paced Learning with Diversity
Authors: Lu Jiang, Deyu Meng, Shoou-I Yu, Zhenzhong Lan, Shiguang Shan, Alexander Hauptmann
NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate that our method significantly outperforms the conventional SPL on three real-world datasets. Specifically, SPLD achieves the best MAP so far reported in literature on the Hollywood2 and Olympic Sports datasets. |
| Researcher Affiliation | Academia | 1School of Computer Science, Carnegie Mellon University 2School of Mathematics and Statistics, Xi an Jiaotong University 3Institute of Computing Technology, Chinese Academy of Sciences |
| Pseudocode | Yes | Algorithm 1: Algorithm for Solving minv E(w, v; λ, γ). Algorithm 2: Algorithm of Self-Paced Learning with Diversity. |
| Open Source Code | Yes | The code is at (http://www.cs.cmu.edu/~lujiang/spld). |
| Open Datasets | Yes | TRECVID MED13Test, which consists of about 32,000 Internet videos... The official test split released by NIST (National Institute of Standards and Technology) is used [15]. Hollywood2 was collected from 69 different Hollywood movies [21]. ... Olympic Sports consists of athletes practicing different sports collected from You Tube [22]. |
| Dataset Splits | Yes | The parameters of all methods are tuned on the same validation set. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts) used for running its experiments. |
| Software Dependencies | No | Lib Linear is used as the solver in Step 4 of Algorithm 2 due to its robust performance on this task. However, no version number for Lib Linear or other software dependencies is provided. |
| Experiment Setup | Yes | By default, the group membership is generated by the spectral clustering, and the number of groups is set to 64. ... Following SPL [6], the self-paced parameters are updated by absolute values of µ1, µ2 (µ1, µ2 ≥ 1) in Step 6 at the end of every iteration. |