Robust Domain Generalisation by Enforcing Distribution Invariance
نویسندگان
چکیده
Many conventional statistical machine learning algorithms generalise poorly if distribution bias exists in the datasets. For example, distribution bias arises in the context of domain generalisation, where knowledge acquired from multiple source domains need to be used in a previously unseen target domains. We propose Elliptical Summary Randomisation (ESRand), an efficient domain generalisation approach that comprises of a randomised kernel and elliptical data summarisation. ESRand learns a domain interdependent projection to a latent subspace that minimises the existing biases to the data while maintaining the functional relationship between domains. In the latent subspace, ellipsoidal summaries replace the samples to enhance the generalisation by further removing bias and noise in the data. Moreover, the summarisation enables large-scale data processing by significantly reducing the size of the data. Through comprehensive analysis, we show that our subspace-based approach outperforms state-of-the-art results on several activity recognition benchmark datasets, while keeping the computational complexity significantly low.
منابع مشابه
Achieving k-anonymity Using Improved Greedy Heuristics for Very Large Relational Databases
Advances in data storage, data collection and inference techniques have enabled the creation of huge databases of personal information. Dissemination of information from such databases even if formally anonymised, creates a serious threat to individual privacy through statistical disclosure. One of the key methods developed to limit statistical disclosure risk is k-anonymity. Several methods ha...
متن کاملSample-oriented Domain Adaptation for Image Classification
Image processing is a method to perform some operations on an image, in order to get an enhanced image or to extract some useful information from it. The conventional image processing algorithms cannot perform well in scenarios where the training images (source domain) that are used to learn the model have a different distribution with test images (target domain). Also, many real world applicat...
متن کاملA Robust Image Denoising Technique in the Contourlet Transform Domain
The contourlet transform has the benefit of efficiently capturing the oriented geometrical structures of images. In this paper, by incorporating the ideas of Stein’s Unbiased Risk Estimator (SURE) approach in Nonsubsampled Contourlet Transform (NSCT) domain, a new image denoising technique is devised. We utilize the characteristics of NSCT coefficients in high and low subbands and apply SURE sh...
متن کاملA Pac Analysis of a Bayesian Estimator Produced as Part of the Esprit Working Group in Neural and Computational Learning, Neurocolt 8556
Bayesian analysis of generalisation can place a prior distribution on the hypotheses and estimate the volume of this space that is consistent with the training data. The larger this volume the greater the conndence in the classiier obtained. The key feature of such estimators is that they provide a posteriori estimates of generalisation based on properties of the hypothesis and the training dat...
متن کامل