A new learning paradigm: Learning using privileged information

By Vladimir Naumovich Vapnik, Akshay Vashist. 2009

891 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Representation & Self-Supervised Learning, Statistical Learning Theory

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 Representation & Self-Supervised Learning

In Statistical Learning Theory

Citations in the corpus

Shaping Visual Representations with Language for Few-Shot Classification (2019, background): “Our work can thus be seen as an instance of the learning using privileged information (LUPI) [31] framework, where richer supervision augments a model during training only.”
Webly Supervised Joint Embedding for Cross-Modal Image-Text Retrieval (2018, methodology): “ for representing tags since tags do not have any semantic context as in the sentences. Our task can also be considered from the perspective of learning with side or privileged information strategies [45, 51], as in our case an additional tag modality is available at training time and we would like ”
Towards the Science of Security and Privacy in Machine Learning (2016, background): “The training data may also include additional features not available at runtime (referred to as privileged information in some settings [27]).”
Learning and Refining of Privileged Information-Based RNNs for Action Recognition from Depth Sequences (2017, background): “This knowledge, also referred to as PI [41], hidden information [50] or side information [54], which can help to provide more explanations in training but will not be available at testing.”
Bayesian representation learning with oracle constraints (2015, background): “Finally, an intuitive connection also exists with Vapnik’s privileged learning framework (Vapnik & Vashist, 2009) where in a supervised setting improved classiﬁers can be learned if privileged information in the form of additional features is present during training time.”
Unifying distillation and privileged information (2015, methodology): “3.1 C OMPARISON TO PRIOR WORK Vapnik & Vashist (2009); Vapnik & Izmailov (2015) offer two strategies to learn using privileged information: similarity control and knowledge transfer.”
SoK: Security and Privacy in Machine Learning (2018, background): “vised learning or a reward function for reinforcement learning. The training data may also include additional features not available at runtime (referred to as privileged information in some settings [27]). As discussed below, the means of collection and validation processes offer another attack sur”
Privileged Modality Distillation for Vessel Border Detection in Intracoronary Imaging (2019, background): “privileged information) to conduce the learning of the target network in the learning phase, but applies the network without the privileged information in the test phase [28].”
Contextual Learning (2020, neutral): “Others sometimes refer to them as privileged information (Vapnik & Vashist, 2009) or side information.”
Better & Faster Large Language Models via Multi-token Prediction (2024, background): “…a successful strategy in various domains, e.g. for learning time series prediction with more distant time steps in the future as auxiliary targets (Vapnik and Vashist, 2009) or for learning from videos with several future frames (Mathieu et al., 2016; Srivastava et al., 2016) or representations…”
Classifier learning with hidden information (2015, methodology): “Hidden information, also referred to as privileged information [1] or side information [2], has been exploited to enhance different learning tasks such as classifier learning [1, 3, 2, 4, 5], feature learning [6], clustering [7], and metric learning [8].”
Adaptive SVM+: Learning with Privileged Information for Domain Adaptation (2017, methodology): “We combine the advantages of the LUPI paradigm [34] and domain adaptation as proposed by Yang et al.”
Randomized Nonlinear Component Analysis (2014, methodology): “Section 5.3 provides a novel algorithm based on RCCA to perform learning using privileged information (Vapnik & Vashist, 2009).”
Supervised mid-level features for word image representation (2014, background): “In that sense, our work can also be related to works on learning with privileged information [37, 33, 10], where the information available at training time to construct the representations (in our case, character bounding boxes) is not available at test time.”
Privileged Information for Modeling Affect In The Wild (2021, methodology): “Vapnik formally introduced the Learning Using Privileged Information (LUPI) paradigm in [7], [8].”
Towards Knowledgeable Supervised Lifelong Learning Systems (2020, methodology): “…has been defined as variants of the original SVM problem, e.g. ν-SVM (Schölkopf, Smola, Williamson, & Bartlett, 2000) and learning with privileged information (Vapnik & Vashist, 2009), which raises the potential to combine these variants to simultaneously achieve different properties of learning.”