A novel approach for unsupervised domain adaptation for neural networks is proposed. It relies on metricbased regularization of the learning process. The metric-based regularization aims at domain-invariant latent feature representations by means of maximizing the similarity between domain-specific activation distributions. The proposed metric results from modifying an integral probability metric such that it becomes less translation-sensitive on a polynomial function space. The metric has an intuitive interpretation in the dual space as the sum of differences of higher order central moments of the corresponding activation distributions. Under appropriate assumptions on the input distributions, error minimization is proven for the continuous case. As demonstrated by an analysis of standard benchmark experiments for sentiment analysis, object recognition and digit recognition, the outlined approach is robust regarding parameter changes and achieves higher classification accuracies than comparable approaches. 1

Multivariate calibration models often fail to extrapolate beyond the calibration samples because of changes associated with the instrumental response, environmental condition, or sample matrix. Most of the current methods used to adapt a source calibration model to a target domain exclusively apply to calibration transfer between similar analytical devices, while generic methods for calibration-model adaptation are largely missing. To fill this gap, we here introduce domain-invariant partial-least-squares (di-PLS) regression, which extends ordinary PLS by a domain regularizer in order to align the source and target distributions in the latent-variable space. We show that a domain-invariant weight vector can be derived in closed form, which allows the integration of (partially) labeled data from the source and target domains as well as entirely unlabeled data from the latter. We test our approach on a simulated data set where the aim is to desensitize a source calibration model to an unknown interfering agent in the target domain (i.e., unsupervised model adaptation). In addition, we demonstrate unsupervised, semisupervised, and supervised model adaptation by di-PLS on two real-world near-infrared (NIR) spectroscopic data sets.

Generative Adversarial Networks have surprising ability for generating sharp and realistic images, though they are known to suffer from the so-called mode collapse problem. In this paper, we propose a new GAN variant called Mixture Density GAN that while being capable of generating high-quality images, overcomes this problem by encouraging the Discriminator to form clusters in its embedding space, which in turn leads the Generator to exploit these and discover different modes in the data. This is achieved by positioning Gaussian density functions in the corners of a simplex, using the resulting Gaussian mixture as a likelihood function over discriminator embeddings, and formulating an objective function for GAN training that is based on these likelihoods. We demonstrate empirically (1) the quality of the generated images in Mixture Density GAN and their strong similarity to real images, as measured by the Fréchet Inception Distance (FID), which compares very favourably with state-of-the-art methods, and (2) the ability to avoid mode collapse and discover all data modes.

The main challenges are discussed together with the lessons learned from past and ongoing research along the development cycle of machine learning systems. This will be done by taking into account intrinsic conditions of nowadays deep learning models, data and software quality issues and human-centered artificial intelligence (AI) postulates, including confidentiality and ethical aspects. The analysis outlines a fundamental theory-practice gap which superimposes the challenges of AI system engineering at the level of data quality assurance, model building, software engineering and deployment. The aim of this paper is to pinpoint research topics to explore approaches to address these challenges.

No abstract