Decision Forests: A Unified Framework for Classification, Regression, Density Estimation, Manifold Learning and Semi-Supervised Learning

By Antonio Criminisi, Jamie Shotton, Ender Konukoglu. 2012

967 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Probabilistic & Bayesian Methods

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

In Probabilistic & Bayesian Methods

Citations in the corpus

Fast Edge Detection Using Structured Forests (2014, methodology): “Edge detection: Numerous papers have been written on edge detection over the past 50 years.”
Playing for Benchmarks (2017, methodology): “To invert these black-box transformations, we have initially considered recording all input and output buffers in the pipeline and learning a non-parametric mapping from output to input via random forests [11].”
Evolutionary Hough Games for coherent object detection (2012, neutral): “A Hough Forest is an ensemble T of binary decision trees, similar to the popular concept of randomized decision trees [11–13].”
Hyperparameters and tuning strategies for random forest (2018, background): “RF has been the topic of several reviews in the last few years including our own review (Boulesteix et al., 2012b) and others (Criminisi et al., 2012; Ziegler and König, 2014; Biau and Scornet, 2016; Belgiu and Drăguţ, 2016).”
Learning to Detect Good 3D Keypoints (2017, methodology): “The latter is realized by leveraging on the ability of Random Forests to handle multi-class classification seamlessly, thereby empowering the learned detector with the capacity of selecting the optimal support size to compute the chosen descriptor and increasing the quantity of correct correspondenc”
Combining embedded accelerometers with computer vision for recognizing food preparation activities (2013, background): “These characteristics make this classiﬁer favorable compared to the popular support vector machine (SVM) [3].”
3D Deep Learning on Medical Images: A Review (2020, methodology): “Historical methods for automated classification of images involves extensive rule-based algorithms or manual feature handcrafting [16–21], which are time-consuming, have poor generalization capacity and require domain knowledge.”
RisQ: Recognizing Smoking Gestures with Inertial Sensors on a Wristband (2014, methodology): “We then use a supervised classifier, called Random Forests [5, 7], that outputs the probability for the type of gesture (“smoking”, “eating” or “other” gesture) based on the extracted features for each segment individually.”
ITK-SNAP: an interactive tool for semi-automatic segmentation of multi-modality biomedical images (2016, methodology): “Multiple ways of obtaining these foreground/background probabilities are available, including thresholding of a given single modality; Gaussian mixture model-based clustering; and supervised classification using Random Forests [1], [2], which is the focus of this paper.”
Indoor Scene Understanding with RGB-D Images: Bottom-up Segmentation, Object Detection and Semantic Segmentation (2015, methodology): “We experiment with random decision tree forests (Breiman 2001; Criminisi et al. 2012) (RF), and additive kernel (Maji et al.”
Spacetime Forests with Complementary Features for Dynamic Scene Recognition (2013, background): “Bootstrapping in the training stage allows maximum training efficiency [6] and avoids overfitting [2] to generalize well in practice.”
Classifier adaptation at prediction time (2015, methodology): “, convolutional networks [1], random forests [6] or one-versus-rest support vector machines [14] with Platt scaling [19].”
Random forest-based similarity measures for multi-modal classification of Alzheimer's disease (2013, positive): “Criminisi et al. (2012), for example, describes this and other extensions to the original random forests algorithm, and presents a unified model of random decision forests for classification, regression, density estimation, manifold learning, and semi-supervised learning.”
SemiContour: A Semi-supervised Learning Approach for Contour Detection (2016, methodology): “Intuitively, minimizing G is to find a separating line in the k-th feature dimension (several feature dimensions can be used together to define a separating hyperplane [11]) to split D̃i in the whole feature space into the left and right subtrees, so that pj on both sides are maximized [6].”
Structured Forests for Fast Edge Detection (2013, background): “Both are computationally efficient and effective in practice [7].”