Detecting and analyzing various defect types in semiconductor materials is an important prerequisite for understanding the underlying mechanisms as well as tailoring the production processes. Analysis of microscopy images that reveal defects typically requires image analysis tasks such as segmentation and object detection. With the permanently increasing amount of data that is produced by experiments, handling these tasks manually becomes more and more impossible. In this work, we combine various image analysis and data mining techniques for creating a robust and accurate, automated image analysis pipeline. This allows for extracting the type and position of all defects in a microscopy image of a KOH-etched 4H-SiC wafer that was stitched together from approximately 40,000 individual images.

Text simplification refers to the process of increasing the comprehensibility of texts. Automatic text simplification models are most commonly evaluated by experts or crowdworkers instead of the primary target groups of simplified texts, such as persons with intellectual disabilities. We conducted an evaluation study of text comprehensibility including participants with and without intellectual disabilities reading unsimplified, automatically and manually simplified German texts on a tablet computer. We explored four different approaches to measuring comprehensibility: multiple-choice comprehension questions, perceived difficulty ratings, response time, and reading speed. The results revealed significant variations in these measurements, depending on the reader group and whether the text had undergone automatic or manual simplification. For the target group of persons with intellectual disabilities, comprehension questions emerged as the most reliable measure, while analyzing reading speed provided valuable insights into participants' reading behavior.

It is becoming increasingly common in regression to train neural networks that model the entire distribution even if only the mean is required for prediction. This additional modeling often comes with performance gain and the reasons behind the improvement are not fully known. This paper investigates a recent approach to regression, the Histogram Loss, which involves learning the conditional distribution of the target variable by minimizing the cross-entropy between a target distribution and a flexible histogram prediction. We design theoretical and empirical analyses to determine why and when this performance gain appears, and how different components of the loss contribute to it. Our results suggest that the benefits of learning distributions in this setup come from improvements in optimization rather than learning a better representation. We then demonstrate the viability of the Histogram Loss in common deep learning applications without a need for costly hyperparameter tuning.

We consider a statistical version of curriculum learning (CL) in a parametric prediction setting. The learner is required to estimate a target parameter vector, and can adaptively collect samples from either the target model, or other source models that are similar to the target model, but less noisy. We consider three types of learners, depending on the level of side-information they receive. The first two, referred to as strong/weak-oracle learners, receive high/low degrees of information about the models, and use these to learn. The third, a fully adaptive learner, estimates the target parameter vector without any prior information. In the single source case, we propose an elimination learning method, whose risk matches that of a strong-oracle learner. In the multiple source case, we advocate that the risk of the weak-oracle learner is a realistic benchmark for the risk of adaptive learners. We develop an adaptive multiple elimination-rounds CL algorithm, and characterize instance-dependent conditions for its risk to match that of the weak-oracle learner. We consider instance-dependent minimax lower bounds, and discuss the challenges associated with defining the class of instances for the bound. We derive two minimax lower bounds, and determine the conditions under which the performance weak-oracle learner is minimax optimal.

The combination of lightly supervised pre-training and online fine-tuning has played a key role in recent AI developments. These new learning pipelines call for new theoretical frameworks. In this paper, we formalize core aspects of weakly supervised and active learning with a simple problem: the estimation of the mode of a distribution using partial feedback. We show how entropy coding allows for optimal information acquisition from partial feedback, develop coarse sufficient statistics for mode identification, and adapt bandit algorithms to our new setting. Finally, we combine those contributions into a statistically and computationally efficient solution to our problem.

Covariate adjustment is a ubiquitous method used to estimate the average treatment effect (ATE) from observational data. Assuming a known graphical structure of the data generating model, recent results give graphical criteria for optimal adjustment, which enables efficient estimation of the ATE. However, graphical approaches are challenging for high-dimensional and complex data, and it is not straightforward to specify a meaningful graphical model of non-Euclidean data such as texts. We propose an general framework that accommodates adjustment for any subset of information expressed by the covariates. We generalize prior works and leverage these results to identify the optimal covariate information for efficient adjustment. This information is minimally sufficient for prediction of the outcome conditionally on treatment. Based on our theoretical results, we propose the Debiased Outcome-adapted Propensity Estimator (DOPE) for efficient estimation of the ATE, and we provide asymptotic results for the DOPE under general conditions. Compared to the augmented inverse propensity weighted (AIPW) estimator, the DOPE can retain its efficiency even when the covariates are highly predictive of treatment. We illustrate this with a single-index model, and with an implementation of the DOPE based on neural networks, we demonstrate its performance on simulated and real data. Our results show that the DOPE provides an efficient and robust methodology for ATE estimation in various observational settings.

Common kernel ridge regression is expensive in memory allocation and computation time. This paper addresses low rank approximations and surrogates for kernel ridge regression, which bridge these difficulties. The fundamental contribution of the paper is a lower bound on the rank of the low dimensional approximation, which is required such that the prediction power remains reliable. The bound relates the effective dimension with the largest statistical leverage score. We characterize the effective dimension and its growth behavior with respect to the regularization parameter by involving the regularity of the kernel. This growth is demonstrated to be asymptotically logarithmic for suitably chosen kernels, justifying low-rank approximations as the Nystr\"om method.

There are several applications of stochastic optimization where one can benefit from a robust estimate of the gradient. For example, domains such as distributed learning with corrupted nodes, the presence of large outliers in the training data, learning under privacy constraints, or even heavy-tailed noise due to the dynamics of the algorithm itself. Here we study SGD with robust gradient estimators based on estimating the median. We first consider computing the median gradient across samples, and show that the resulting method can converge even under heavy-tailed, state-dependent noise. We then derive iterative methods based on the stochastic proximal point method for computing the geometric median and generalizations thereof. Finally we propose an algorithm estimating the median gradient across iterations, and find that several well known methods - in particular different forms of clipping - are particular cases of this framework.

The Poisson process, especially the nonhomogeneous Poisson process (NHPP), is an essentially important counting process with numerous real-world applications. Up to date, almost all works in the literature have been on the estimation of NHPPs with infinite data using non-data driven binning methods. In this paper, we formulate the problem of estimation of NHPPs from finite and limited data as a learning generalization problem. We mathematically show that while binning methods are essential for the estimation of NHPPs, they pose a threat of overfitting when the amount of data is limited. We propose a framework for regularized learning of NHPPs with two new adaptive and data-driven binning methods that help to remove the ad-hoc tuning of binning parameters. Our methods are experimentally tested on synthetic and real-world datasets and the results show their effectiveness.

We consider the problem of Sparse Principal Component Analysis (PCA) when the ratio $d/n \rightarrow c > 0$. There has been a lot of work on optimal rates on sparse PCA in the offline setting, where all the data is available for multiple passes. In contrast, when the population eigenvector is $s$-sparse, streaming algorithms that have $O(d)$ storage and $O(nd)$ time complexity either typically require strong initialization conditions or have a suboptimal error. We show that a simple algorithm that thresholds and renormalizes the output of Oja's algorithm (the Oja vector) obtains a near-optimal error rate. This is very surprising because, without thresholding, the Oja vector has a large error. Our analysis centers around bounding the entries of the unnormalized Oja vector, which involves the projection of a product of independent random matrices on a random initial vector. This is nontrivial and novel since previous analyses of Oja's algorithm and matrix products have been done when the trace of the population covariance matrix is bounded while in our setting, this quantity can be as large as $n$.