On feature combination for multiclass object classification

By Peter Gehler, Sebastian Nowozin. IEEE International Conference on Computer Vision, 2009

926 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/ba0548583a5ab3dca551f60e30f85ea42b2a4873

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Fusing Multi-Stream Deep Networks for Video Classification (2015, result): “Kernel average fusion shows slightly better results than MKL, which is consistent with the observations in several previous studies like [12].”
A survey of depth and inertial sensor fusion for human action recognition (2015, methodology): “(Feature-level fusion) [47] Multiple kernel learning (MKL) strategy [28] was used for combining various modalities for action recognition.”
Hierarchical Regularization Cascade for Joint Learning (2013, methodology): “In the original multi-class setting addressed in (Gehler & Nowozin, 2009) our results compare to their state-of-the-art results both for 30 training images (78.1”
Large Scale Visual Recognition (2012, methodology): “In the current state of the art, multiple descriptors and kernels are combined using either ad hoc or multiple kernel learning approaches [78, 71, 79, 80].”
Hierarchical Coding Vectors for Scene Level Land-Use Classification (2016, methodology): “We used the multiple kernel learning [31] method with the average kernel to combine HCV with FV.”
Visual classification with multi-task joint sparse representation (2010, background): “We show that the kernel-view extensions of our algorithm are competitive to the state-of-the-art multiple kernel learning methods [7][14][21] for object recognition.”
A Binary Classification Framework for Two-Stage Multiple Kernel Learning (2012, methodology): “While LP-β is similar in spirit to multiple-kernel learning, it is not a true kernel learning algorithm as it does not produce a ker- 6C = 1000 is the best setting for the one-stage MKL algorithms (Gehler & Nowozin, 2009) 7We take the results for LP-β and MKL from (Gehler & Nowozin, 2009).”
Semi-Supervised Manifold Learning Based Multigraph Fusion for High-Resolution Remote Sensing Image Classification (2014, methodology): “The following six methods are compared with SSM-MF: Feature Concatenation (FC), Best Single Feature (BSF), Best Single Feature Embedding (BSFE), MFC [3] and two kernel fusion methods proposed in [18]: Average Kernel Fusion (AKF) and Product Kernel Fusion (PKF).”
Information Bottleneck Learning Using Privileged Information for Visual Recognition (2016, background): “Indeed, our problem is more related to multi-view and multi-task learning [9, 40, 14, 28, 45].”
Ontology-Based Semantic Image Segmentation Using Mixture Models and Multiple CRFs (2016, background): “shape or texture) strongly correlates with the class that an image belongs to [24] and [25], proper feature weighing”
TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 1 ModDrop (2018, neutral): “At the same time, as shown by [41], simple additive or multiplicative averaging of kernels may reach the same level of performance while being orders of magnitude faster.”
Kernel Descriptors for Visual Recognition (2010, background): “Match kernels defined over various pixel attributes provide a unified way to generate a rich, diverse visual feature set, which has been shown to be very successful to boost recognition accuracy [6].”
Kervolutional Neural Networks (2019, background): “This provides a unified way to generate a rich, diverse visual feature set [15].”
Semantic Kernel Forests from Multiple Taxonomies (2012, background): “, [31, 32]) and are frequently employed in practice as a principled way to combine feature types.”