Learning Categories From Few Examples With Multi Model Knowledge Transfer

By Tatiana Tommasi, Francesco Orabona, Barbara Caputo. 2014

174 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning, Optimization & Training

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 Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

In Optimization & Training

Citations in the corpus

Growing a Brain: Fine-Tuning by Increasing Model Capacity (2017, background): “As we will show later in our experiments, this signiﬁcantly improves the ability to transfer knowledge to target tasks, particularly when fewer target examples are provided [43].”
Learning Like a Child: Fast Novel Visual Concept Learning from Sentence Descriptions of Images (2015, methodology): “Another related task is one-shot learning task of new categories [16, 29, 53].”
Data-Dependent Stability of Stochastic Gradient Descent (2017, background): “This is related to TL where one transfers from pre-trained models (Kuzborskij & Orabona, 2016; Tommasi et al., 2014; Pentina & Lampert, 2014; Ben-David & Urner, 2013), especially popular in deep learning due to its data-demanding nature (Galanti et al.”
Rectify Heterogeneous Models with Semantic Mapping (2018, background): “1 can also be transformed to learn a model bias ∆W based on the existing W0, and then predicts with W0 + ∆W (Tommasi et al., 2014).”
A Comprehensive Survey on Test-Time Adaptation Under Distribution Shifts (2023, unknown): “Shallow HTL methods [54], [55], [56], [57], [58], [59] generally assume that the optimal target hypothesis is closely related to these source hypotheses.”
Model recommendation: Generating object detectors from few samples (2015, background): “For instance, the source models in transfer learning are usually well-trained categorical classifiers [49, 4].”
Fast Generalized Distillation for Semi-Supervised Domain Adaptation (2017, background): “As suggested by (Tommasi, Orabona, and Caputo 2014), learning from multiple related sources simultaneously can significantly improve the performance of the target model.”
Do We Really Need to Access the Source Data? Source Hypothesis Transfer for Unsupervised Domain Adaptation (2020, background): “Conventional HTL always requires labeled data in the target domain (Tommasi et al., 2013) or multiple hypotheses from different source domains (Mansour et al., 2009), which however cannot be applied to unsupervised DA.”
Heterogeneous Few-Shot Model Rectification With Semantic Mapping (2020, methodology): “(2) equals learning D W and predicting with W 0 þ D W [8].”
Learning to Learn: Model Regression Networks for Easy Small Sample Learning (2016, methodology): “In addition, adaptive SVM and its variants present SVM-based model adaptation by combining classifiers learned on related categories [47, 102, 23, 2, 3, 53, 88, 97].”
Domain Adaptation by Mixture of Alignments of Second-or Higher-Order Scatter Tensors (2016, background): “Therefore, one has to address three questions: what to transfer, how to transfer, and when to transfer [39].”
Phase Transitions in Transfer Learning for High-Dimensional Perceptrons (2021, unknown): “It has been attracting significant interest in recent literature [4]–[9], [11], [12].”
Towards Knowledgeable Supervised Lifelong Learning Systems (2020, background): “Most research in hypothesis transfer learning with SVM has proposed to transfer knowledge from source hypotheses as a whole, without distinguishing fragments of knowledge that can be potentially more useful for a target task (Yang et al., 2007; Aytar & Zisserman, 2011; Tommasi et al., 2014; Kuzborsk”
Transfer Learning via Learning to Transfer (2018, background): “Though limited parameter-based algorithms (Yang et al., 2007a; Tommasi et al., 2014) can transfer across domains in heterogeneous label spaces, they can only handle binary classiﬁcation problems.”