Results 1 
8 of
8
Classifier Adaptation at Prediction Time
"... Classifiers for object categorization are usually evaluated by their accuracy on a set of i.i.d. test examples. This provides us with an estimate of the expected error when applying the classifiers to a single new image. In real application, however, classifiers are rarely only used for a single ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
(Show Context)
Classifiers for object categorization are usually evaluated by their accuracy on a set of i.i.d. test examples. This provides us with an estimate of the expected error when applying the classifiers to a single new image. In real application, however, classifiers are rarely only used for a single image and then discarded. Instead, they are applied sequentially to many images, and these are typically not i.i.d. samples from a fixed data distribution, but they carry dependencies and their class distribution varies over time. In this work, we argue that the phenomenon of correlated data at prediction time is not a nuisance, but a blessing in disguise. We describe a probabilistic method for adapting classifiers at prediction time without having to retrain them. We also introduce a framework for creating realistically distributed image sequences, which offers a way to benchmark classifier adaptation methods, such as the one we propose. Experiments on the ILSVRC2010 and ILSVRC2012 datasets show that adapting object classification systems at prediction time can significantly reduce their error rate, even with no additional human feedback. 1.
Lost in the Past: Recognizing Locations Over Large Time Lags
"... Would it be possible to automatically associate ancient pictures to modern ones and create fancy cultural heritage city maps? We introduce here the task of recognizing the location depicted in an old photo given modern annotated images collected from the Internet. We present an extensive analysis o ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
(Show Context)
Would it be possible to automatically associate ancient pictures to modern ones and create fancy cultural heritage city maps? We introduce here the task of recognizing the location depicted in an old photo given modern annotated images collected from the Internet. We present an extensive analysis on different features, looking for the most discriminative and most robust to the image variability induced by large time lags. Moreover, we show that the described task benefits from domain adaptation. We show that using existing domain adaptation methods it is possible to obtain promising results in both location recognition and interactive location retrieval. 1
Published as a conference paper at ICLR 2015 A UNIFIED PERSPECTIVE ON MULTIDOMAIN AND MULTITASK LEARNING
"... In this paper, we provide a new neuralnetwork based perspective on multitask learning (MTL) and multidomain learning (MDL). By introducing the concept of a semantic descriptor, this framework unifies MDL and MTL as well as encompassing various classic and recent MTL/MDL algorithms by interpretin ..."
Abstract
 Add to MetaCart
In this paper, we provide a new neuralnetwork based perspective on multitask learning (MTL) and multidomain learning (MDL). By introducing the concept of a semantic descriptor, this framework unifies MDL and MTL as well as encompassing various classic and recent MTL/MDL algorithms by interpreting them as different ways of constructing semantic descriptors. Our interpretation provides an alternative pipeline for zeroshot learning (ZSL), where a model for a novel class can be constructed without training data. Moreover, it leads to a new and practically relevant problem setting of zeroshot domain adaptation (ZSDA), which is the analogous to ZSL but for novel domains: A model for an unseen domain can be generated by its semantic descriptor. Experiments across this range of problems demonstrate that our framework outperforms a variety of alternatives. 1
YANG AND HOSPEDALES: ZSDA VIA KERNEL REGRESSION ON THE GRASSMANNIAN 1 ZeroShot Domain Adaptation via Kernel Regression on the Grassmannian
"... Most visual recognition methods implicitly assume the data distribution remains unchanged from training to testing. However, in practice domain shift often exists, where realworld factors such as lighting and sensor type change between train and test, and classifiers do not generalise from source ..."
Abstract
 Add to MetaCart
(Show Context)
Most visual recognition methods implicitly assume the data distribution remains unchanged from training to testing. However, in practice domain shift often exists, where realworld factors such as lighting and sensor type change between train and test, and classifiers do not generalise from source to target domains. It is impractical to train separate models for all possible situations because collecting and labelling the data is expensive. Domain adaptation algorithms aim to ameliorate domain shift, allowing a model trained on a source to perform well on a different target domain. However, even for the setting of unsupervised domain adaptation, where the target domain is unlabelled, collecting data for every possible target domain is still costly. In this paper, we propose a new domain adaptation method that has no need to access either data or labels of the target domain when it can be described by a parametrised vector and there exits several related source domains within the same parametric space. It greatly reduces the burden of data collection and annotation, and our experiments show some promising results. 1
Predicting the Future Behavior of a TimeVarying Probability Distribution
"... We study the problem of predicting the future, though only in the probabilistic sense of estimating a future state of a timevarying probability distribution. This is not only an interesting academic problem, but solving this extrapolation problem also has many practical application, e.g. for train ..."
Abstract
 Add to MetaCart
(Show Context)
We study the problem of predicting the future, though only in the probabilistic sense of estimating a future state of a timevarying probability distribution. This is not only an interesting academic problem, but solving this extrapolation problem also has many practical application, e.g. for training classifiers that have to operate under timevarying conditions. Our main contribution is a method for predicting the next step of the timevarying distribution from a given sequence of sample sets from earlier time steps. For this we rely on two recent machine learning techniques: embedding probability distributions into a reproducing kernel Hilbert space, and learning operators by vectorvalued regression. We illustrate the working principles and the practical usefulness of our method by experiments on synthetic and real data. We also highlight an exemplary application: training a classifier in a domain adaptation setting without having access to examples from the test time distribution at training time. 1.
Adaptation Based on Generalized Discrepancy
"... We present a new algorithm for domain adaptation improving upon a discrepancy minimization algorithm, (DM), previously shown to outperform a number of algorithms for this problem. Unlike many previously proposed solutions for domain adaptation, our algorithm does not consist of a fixed reweighting o ..."
Abstract
 Add to MetaCart
(Show Context)
We present a new algorithm for domain adaptation improving upon a discrepancy minimization algorithm, (DM), previously shown to outperform a number of algorithms for this problem. Unlike many previously proposed solutions for domain adaptation, our algorithm does not consist of a fixed reweighting of the losses over the training sample. Instead, the reweighting depends on the hypothesis sought. The algorithm is derived from a less conservative notion of discrepancy than the DM algorithm called generalized discrepancy. We present a detailed description of our algorithm and show that it can be formulated as a convex optimization problem. We also give a detailed theoretical analysis of its learning guarantees which helps us select its parameters. Finally, we report the results of experiments demonstrating that it improves upon discrepancy minimization in several tasks.
Adaptation Algorithm and Theory Based on Generalized Discrepancy
"... We present a new algorithm for domain adaptation improving upon the discrepancy minimization algorithm (DM), which was previously shown to outperform a number of popular algorithms designed for this task. Unlike most previous approaches adopted for domain adaptation, our algorithm does not consist ..."
Abstract
 Add to MetaCart
(Show Context)
We present a new algorithm for domain adaptation improving upon the discrepancy minimization algorithm (DM), which was previously shown to outperform a number of popular algorithms designed for this task. Unlike most previous approaches adopted for domain adaptation, our algorithm does not consist of a fixed reweighting of the losses over the training sample. Instead, it uses a reweighting that depends on the hypothesis considered and is based on the minimization of a new measure of generalized discrepancy. We give a detailed description of our algorithm and show that it can be formulated as a convex optimization problem. We also present a detailed theoretical analysis of its learning guarantees, which helps us select its parameters. Finally, we report the results of experiments demonstrating that it improves upon the DM algorithm in several tasks. 1.
Blind Domain Adaptation: An RKHS Approach
"... We study the problem of domain adaptation: our goal is to learn a classifier, but the data distribution at training time (source) differs from the data distribution at prediction time (target). In contrast to existing work, we do not assume any samples from the target distribution to be available al ..."
Abstract
 Add to MetaCart
(Show Context)
We study the problem of domain adaptation: our goal is to learn a classifier, but the data distribution at training time (source) differs from the data distribution at prediction time (target). In contrast to existing work, we do not assume any samples from the target distribution to be available already at training time, not even unlabeled ones. Instead, we assume that the distribution mismatch is due to an underlying timeevolution of the data distribution, and that we have access to sample sets from more than one earlier time steps. Our main contribution is a method for learning an operator that can extrapolate the dynamics of the data distribution. For this we rely on two recent techniques: the embedding of probability distributions into a reproducing kernel Hilbert space, and vectorvalued regression. By extrapolating the learned dynamics into the future, we obtain an estimate of the target distribution, based on which we can either directly learn a classifier for the target situation, or create a new sample set. Experiments on synthetics and real data show the effectiveness of our approach. 1