Reframing Continuous Input Attributes

Abstract

Reuse of learnt knowledge is of critical importance in the majority of knowledge-intensive application areas, particularly because the operating context can be expected to vary from training to deployment. Dataset shift is a crucial example of this where training and testing datasets follow different distributions. However, most of the existing dataset shift solving algorithms need costly retraining operation and are not suitable to use the existing model. In this paper, we propose a new approach called reframing to handle dataset shift. The main objective of reframing is to build a model once and make it workable without retraining. We propose two efficient reframing algorithms to learn the optimal shift parameter values using only a small amount of labelled data available in the deployment. Thus, they can transform the shifted input attributes with the optimal parameter values and use the same existing model in several deployment environments without retraining. We have addressed supervised learning tasks both for classification and regression. Extensive experimental results demonstrate the efficiency and effectiveness of our approach compared to the existing solutions. In particular, we report the existence of dataset shift in two real-life datasets. These real-life unknown shifts can also be accurately modeled by our algorithms.

Versatile Decision Trees for Learning Over Multiple Contexts

“…Such degradation can be seen as a transformation function in the sensor outputs that causes a covariate observation shift. Previous authors dealt with covariate observation shifts by finding a transformation function Φ to correct the deployment data [1]. Once transformed or 'unshifted' using Φ, the deployment data is given as input to the model learned in the training context.…”

Reframing in context: A systematic approach for model reuse in machine learning

“…One way to confirm a hypothetical shift is if it improves a model's performance on some deployment data [3].…”

“…Some of the contexts seen in previous sections may be composed of several parameters (e.g., the coefficient values in some input reframing approaches [3], so θ has twice as many parameters as attributes). In the multidimensional datamart example [46] the operating context is an OLAP cube, and θ is not only composed of many parameters but they are also discrete.…”

“…In the input shift example [3] the context affects the input attributes, leading to an output change through the model. As the relation between the inputs and outputs is model-dependent, it is not possible, in general, to derive the curves analytically.…”

“…To illustrate, input attribute values can be shifted from source to deployment [3]. A model is trained in source and deployed over several different contexts by transforming the input attributes to the appropriate values using only few labelled deployment data.…”