Searcharxiv⌕ Search

arXiv subjects

Michael Klar

Publications and source records attributed to Michael Klar.

5 recordsLinked to original sources

Predictive maintenance in manufacturing: The operational value of continual machine learning

Predictive maintenance in high-volume manufacturing often faces a domain-incremental learning problem: as new product configurations enter production, prediction models must learn from newly available data while preserving performance on earlier configurations that remain in operation. In this paper, we empirically study the operational value of continual machine learning methods for addressing this challenge in predictive maintenance, using real-world maintenance data from Bosch. The setting at Bosch covers 25 product configurations over five years, which creates a practical challenge for predictive maintenance models, as such models must be updated with data from newly introduced configurations without reducing performance on configurations learned earlier, a phenomenon known as "forgetting". At the same time, retaining all historical sensor data and repeatedly retraining models is often infeasible in high-volume manufacturing since industrial machines generate too large data volumes. To address this challenge, we combine continual machine learning with robust optimization to support predictive maintenance. We then compare continual machine learning with a broad set of baselines, including fine-tuning, domain-adaptation methods, online learning methods, and configuration-specific models. Across experiments, continual machine learning methods achieve stronger predictive performance and lower maintenance costs than the baselines, with a reduction of cost-related indicators of more than 40 % relative to the baselines. Our results demonstrate the operational value of continual machine learning for decision support in maintenance when product portfolios grow over time and historical manufacturing data cannot be fully retained.

cs.LG↗

Prediction-powered Neural Architecture Search

Evaluating candidate architectures in neural architecture search (NAS) faces an inherent trade-off: on the one hand, reliable performance labels are limited because training and evaluating architectures is costly; on the other hand, zero-cost proxies (ZCPs) are cheap to compute at large scale but can be noisy. Yet, how to effectively combine these two sources of supervision remains unclear. In this paper, we propose PPNAS, a novel prediction-powered inference (PPI) approach for NAS. PPNAS fuses (1) a small set of architectures with observed performance labels and (2) a large set of architectures with ZCP information. To combine these two sources of supervision, PPNAS exploits the ordinal information provided by ZCPs to construct additional pairwise ranking supervision, while PPI debiases systematic discrepancies between ZCP-based and true performance rankings. We evaluate PPNAS in end-to-end predictor-based NAS, where it achieves state-of-the-art under limited evaluation budgets. To the best of our knowledge, PPNAS is the first prediction-powered approach for label-efficient NAS.

cs.LG↗

ORTHOBO: Orthogonal Bayesian Hyperparameter Optimization

Bayesian optimization is widely used for hyperparameter optimization when model evaluations are expensive; however, noisy acquisition estimates can lead to unstable decisions. We identify acquisition estimation noise as a failure mode that was previously overlooked: even when the surrogate model and acquisition target are correctly specified, finite-sample Monte Carlo error can perturb acquisition values. This can, in turn, flip candidate rankings and lead to suboptimal BO decisions. As a remedy, we aim at variance reduction and propose an orthogonal acquisition estimator that subtracts an optimally weighted score-function control variate, which yields an acquisition residual orthogonal to posterior score directions and which thus reduces Monte Carlo variance. We further introduce OrthoBO: a Bayesian optimization framework that combines our orthogonal acquisition estimator with ensemble surrogates and an outer log transformation. We show theoretically that our estimator preserves the target, leads to variance reduction, and improves pairwise ranking stability. We further verify the theoretical properties of OrthoBO through numerical experiments where our framework reduces acquisition estimation variance, stabilizes candidate rankings, and achieves strong performance. We also demonstrate the downstream utility of OrthoBO in hyperparameter optimization for neural network training and fine-tuning.

cs.LG↗

Robust Learning Under Label Noise With Iterative Noise-Filtering

We consider the problem of training a model under the presence of label noise. Current approaches identify samples with potentially incorrect labels and reduce their influence on the learning process by either assigning lower weights to them or completely removing them from the training set. In the first case the model however still learns from noisy labels; in the latter approach, good training data can be lost. In this paper, we propose an iterative semi-supervised mechanism for robust learning which excludes noisy labels but is still able to learn from the corresponding samples. To this end, we add an unsupervised loss term that also serves as a regularizer against the remaining label noise. We evaluate our approach on common classification tasks with different noise ratios. Our robust models outperform the state-of-the-art methods by a large margin. Especially for very large noise ratios, we achieve up to 20 % absolute improvement compared to the previous best model.

cs.LG↗

Anomaly Detection With Multiple-Hypotheses Predictions

In one-class-learning tasks, only the normal case (foreground) can be modeled with data, whereas the variation of all possible anomalies is too erratic to be described by samples. Thus, due to the lack of representative data, the wide-spread discriminative approaches cannot cover such learning tasks, and rather generative models, which attempt to learn the input density of the foreground, are used. However, generative models suffer from a large input dimensionality (as in images) and are typically inefficient learners. We propose to learn the data distribution of the foreground more efficiently with a multi-hypotheses autoencoder. Moreover, the model is criticized by a discriminator, which prevents artificial data modes not supported by data, and enforces diversity across hypotheses. Our multiple-hypothesesbased anomaly detection framework allows the reliable identification of out-of-distribution samples. For anomaly detection on CIFAR-10, it yields up to 3.9% points improvement over previously reported results. On a real anomaly detection task, the approach reduces the error of the baseline models from 6.8% to 1.5%.

cs.CV↗