CPM-Nets: Cross Partial Multi-View Networks

Authors: Changqing Zhang, Zongbo Han, yajie cui, Huazhu Fu, Joey Tianyi Zhou, Qinghua Hu

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 3 Experiments, We conduct experiments on the following datasets: ORL, PIE, Yale B, CUB, Handwritten, Animal, We compared the proposed CPM-Nets with the following methods:, From the results in Fig. 2, we have the following observations:
Researcher Affiliation Collaboration 1College of Intelligence and Computing, Tianjin University, Tianjin, China 2Tianjin Key Lab of Machine Learning, Tianjin, China 3Inception Institute of Artificial Intelligence, Abu Dhabi, UAE 4Institute of High Performance Computing, A*STAR, Singapore
Pseudocode Yes Algorithm 1: Algorithm for CPM-Nets
Open Source Code No The paper does not provide concrete access to source code, nor does it state that the code is publicly available.
Open Datasets Yes ORL 2https://www.cl.cam.ac.uk/research/dtg/attarchive/facedatabase.html, PIE 3http://www.cs.cmu.edu/afs/cs/project/PIE/Multi Pie/Multi-Pie/Home.html, Handwritten 4https://archive.ics.uci.edu/ml/datasets/Multiple+Features, CUB [38] The dataset contains different categories of birds, where the first 10 categories are used and deep visual features from Goog Le Net and text features using doc2vec [39] are used as two views.
Dataset Splits Yes For all methods, we tune the parameters with 5-fold cross validation.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, processor types, or memory amounts) used for running its experiments.
Software Dependencies No The paper mentions models like Goog Le Net, doc2vec, DECAF, and VGG19, but does not provide specific software dependencies with version numbers for the implementation or experiments.
Experiment Setup Yes For our CPM-Nets, we set the dimensionality (K) of the latent representation from {64, 128, 256} and tune the parameter λ from the set {0.1, 1, 10} for all datasets. We run 10 times for each method to report the mean values and standard deviations. Please refer to the supplementary material for the details of network architectures and parameter settings.