Gaussian Processes for Object Categorization

By Ashish Kapoor, Kristen Grauman, Raquel Urtasun, Trevor Darrell. 2010

213 citations recorded by the source corpus.

Discriminative methods for visual object category recognition are typically non-probabilistic, predicting class labels but not directly providing an estimate of uncertainty. Gaussian Processes (GPs) provide a framework for deriving regression techniques with explicit uncertainty models; we show here how Gaussian Processes with covariance functions defined based on a Pyramid Match Kernel (PMK) can be used for probabilistic object category recognition. Our probabilistic formulation provides a principled way to learn hyperparameters, which we utilize to learn an optimal combination of multiple covariance functions. It also offers confidence estimates at test points, and naturally allows for an active learning paradigm in which points are optimally selected for interactive labeling. We show that with an appropriate combination of kernels a significant boost in classification performance is possible. Further, our experiments indicate the utility of active learning with probabilistic predictive models, especially when the amount of training data labels that may be sought for a category is ultimately very small.

Subfields: Artificial Intelligence

How researchers see this paper

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

In Artificial Intelligence

Citations in the corpus

A review of novelty detection (2014, background): “[90] explore an heuristic measure: the predictive mean divided by the standard deviation, which was proposed in [89] as a combined measure for describing the uncertainty of the estimation.”
Multi-Class Active Learning by Uncertainty Sampling with Diversity Maximization (2015, methodology): “The uncertainty sampling strategy has also been combined with other classifiers, such as the Gaussian process (Kapoor et al. 2010), the K-nearest neighbor classifier (Lindenbaum et al. 2004) and the probabilistic K-nearest neighbor classifier (Jain and Kapoor 2009).”
Introspective perception: Learning to predict failures in vision systems (2016, background): “The concept of introspection as introduced here can be closely related to active learning and incremental learning [4], [5], [6], where uncertainty estimates and model selection steps are used to guide data acquisition and selection.”
Crowdsourcing in Computer Vision (2016, unknown): “The system might select that sample for labeling for which the current classifier has the highest uncertainty over the class label [Tong and Koller, 2002, Kapoor et al., 2010] or has the smallest “peak” in the probability distribution over the class labels [Jain and Kapoor, 2009].”
Heterogeneous Visual Features Fusion via Sparse Multimodal Machine (2013, methodology): “Besides, we also compare our method to three most recent multi-model image classification methods published in computer vision community, including Gaussian process (GP) method [7], LPBoost-β method [5] and LPBoost-B method [5], which have demonstrated state-of-the-art object categorization performa”
Learning to Recognize Objects from Unseen Modalities (2010, methodology): “We did not optimize αm and βm, since it has been shown that averaging the kernels yields similar performance [2, 22].”