ThiNet: A Filter Level Pruning Method for Deep Neural Network Compression
- Jian-Hao Luo, Jianxin Wu, Weiyao Lin
- Computer ScienceIEEE International Conference on Computer Vision
- 20 July 2017
ThiNet is proposed, an efficient and unified framework to simultaneously accelerate and compress CNN models in both training and inference stages, and it is revealed that it needs to prune filters based on statistics information computed from its next layer, not the current layer, which differentiates ThiNet from existing methods.
CENTRIST: A Visual Descriptor for Scene Categorization
- Jianxin Wu, James M. Rehg
- Computer ScienceIEEE Transactions on Pattern Analysis and Machineā¦
- 1 August 2011
CENsus TRansform hISTogram (CENTRIST), a new visual descriptor for recognizing topological places or scene categories, is introduced and is shown to be a holistic representation and has strong generalizability for category recognition.
Ensembling neural networks: Many could be better than all
- Zhi-Hua Zhou, Jianxin Wu, Wei Tang
- Computer ScienceArtificial Intelligence
- 1 May 2002
Exploratory Under-Sampling for Class-Imbalance Learning
- Xu-Ying Liu, Jianxin Wu, Zhi-Hua Zhou
- Computer ScienceIndustrial Conference on Data Mining
- 18 December 2006
Experiments show that the proposed algorithms, BalanceCascade and EasyEnsemble, have better AUC scores than many existing class-imbalance learning methods and have approximately the same training time as that of under-sampling, which trains significantly faster than other methods.
Probabilistic End-To-End Noise Correction for Learning With Noisy Labels
- Kun Yi, Jianxin Wu
- Computer ScienceComputer Vision and Pattern Recognition
- 19 March 2019
An end-to-end framework called PENCIL, which can update both network parameters and label estimations as label distributions and is more general and robust than existing methods and is easy to apply.
Selective Convolutional Descriptor Aggregation for Fine-Grained Image Retrieval
- Xiu-Shen Wei, Jian-Hao Luo, Jianxin Wu, Zhi-Hua Zhou
- Computer ScienceIEEE Transactions on Image Processing
- 18 April 2016
The selective convolutional descriptor aggregation (SCDA) method is proposed, which is unsupervised, using no image label or bounding box annotation, and achieves comparable retrieval results with the state-of-the-art general image retrieval approaches.
Deep Label Distribution Learning With Label Ambiguity
- Bin-Bin Gao, Chao Xing, Chen-Wei Xie, Jianxin Wu, Xin Geng
- Computer ScienceIEEE Transactions on Image Processing
- 6 November 2016
The proposed deep label distribution learning (DLDL) method effectively utilizes the label ambiguity in both feature learning and classifier learning, which help prevent the network from overfitting even when the training set is small.
Beyond the Euclidean distance: Creating effective visual codebooks using the Histogram Intersection Kernel
- Jianxin Wu, James M. Rehg
- Computer ScienceIEEE International Conference on Computer Vision
- 1 September 2009
It is demonstrated that HIK can also be used in an unsupervised manner to significantly improve the generation of visual codebooks and the standard k-median clustering method can be used for visual codebook generation and can act as a compromise between HIK and k-means approaches.
Fast Asymmetric Learning for Cascade Face Detection
- Jianxin Wu, S. Brubaker, M. D. Mullin, James M. Rehg
- Computer ScienceIEEE Transactions on Pattern Analysis and Machineā¦
- 1 March 2008
A linear asymmetric classifier (LAC) is presented, a classifier that explicitly handles the asymmetric learning goal as a well-defined constrained optimization problem and is demonstrated experimentally that LAC results in an improved ensemble classifier performance.
Where am I: Place instance and category recognition using spatial PACT
- Jianxin Wu, James M. Rehg
- Computer ScienceIEEE Conference on Computer Vision and Patternā¦
- 23 June 2008
Spatial PACT is introduced, a new representation for recognizing instances and categories of places or scenes that outperforms the current state-of-the-art in several place and scene recognition, and shape matching datasets.
...
...