Organizations: Department of Computing, Jonkoping University, Jonkoping, Sweden · Department of Information Science, University of Haifa, Haifa, Israel
Manufacturing test flows in high-volume electronics production are typically fixed during product development and executed unchanged on every unit, even as failure patterns and process conditions evolve. This protects quality, but it also imposes unnecessary test cost, while existing data-driven methods mostly optimize static test subsets and neither adapt online to changing defect distributions nor explicitly control escape risk. In this study, we present an adaptive test-selection framework that combines offline minimum-cost diagnostic subset construction using greedy set cover with an online Thompson-sampling multi-armed bandit that switches between full and reduced test plans using a rolling process-stability signal. We evaluate the framework on two printed circuit board assembly stages-Functional Circuit Test and End-of-Line test-covering 28,000 board runs. Offline analysis identified zero-escape reduced plans that cut test time by 18.78% in Functional Circuit Test and 91.57% in End-of-Line testing. Under temporal validation with real concept drift, static reduction produced 110 escaped defects in Functional Circuit Test and 8 in End-of-Line, whereas the adaptive policy reduced escapes to zero by reverting to fuller coverage when instability emerged in practice. These results show that online learning can preserve manufacturing quality while reducing test burden, offering a practical route to adaptive test planning across production domains, and offering both economic and logistics improvement for companies.
Figures & tables
Figure 1: PCBA manufacturing test flow
Figure 2: A schematic view of the multi-arm bandit algorithms and their production tests prediction.
Symbol
Definition
T={1,…,M}
Index set of all available test steps
M
Total number of test steps
ci
Execution cost (mean time in seconds) of test step i
cfull
Total cost of executing all test steps
N
Number of historical units in training dataset
Y∈{0,1}N×M
Binary outcome matrix
Table 1: Summary of the parameters used in the model with their notation and definition.
Figure 3: Schematic flow of the proposed framework.
Symbol
Equation
Explanation
Cred selection rate ( SelRate(π) )
N1t=1∑N1[at=1]
Measures the fraction of boards for which the MAB agent selected the reduced test plan rather than the full plan. Here, at∈{0,1} is the arm selected for board t , and N is the total number of boards processed. A value of 1.0 means the agent always selected the reduced plan, while a value of 0 means the full plan was always used. Higher values indicate greater exploitation of the cost-saving opportunity.
Cost saving (%) ( Saving(π) )
(1−cfullcˉπ)×100,cˉπ=N1t=1∑Ncat
Measures the percentage reduction in mean test time per board relative to executing the full sequence on every board. Here, cat is the cost of the plan selected for board t , and cfull=∑i∈Tci is the cost of the full test sequence. A saving of 0% corresponds to always running the full plan, while the maximum achievable saving corresponds to always running Cred .
Escaped units ( Escaped(π) )
u∈UF∑1{Dπ(yu)(u)=0}
Counts the absolute number of genuinely defective boards that were not detected by the executed test plan and would therefore have proceeded to shipment undetected. Here, DC(u)=maxi∈C(1−yu,i) . A value of zero means all defective boards were caught. Any non-zero value represents a direct quality failure: a defective board that reached the customer undetected.
Empirical escape risk ( R^(π) )
∣UF∣1u∈UF∑1{Dπ(yu)(u)=0}
Expresses the same quantity as a fraction of the total defect population, making it comparable across datasets with different numbers of failing boards. A value of R^(π)=0 indicates complete defect coverage, meaning that all defective boards were detected. This metric is directly comparable to the quality threshold δ , which requires R^(π)≤δ at all times.
Table 2: Evaluation metrics used to assess the policy π .
Test
Cohort
Algorithm
β
Cred (%)
Saving (%)
Escaped
R^(π)
FCT
Training
Baseline ( Cfull )
—
0.0
0.00
0
0.00
Static Cred
—
100.0
18.78
0
0.00
Epsilon-Greedy
10
89.6
16.83
0
0.00
UCB
10
92.8
17.42
0
0.00
Thompson Sampling
10
93.8
17.61
0
0.00
Validation
Baseline ( Cfull )
—
0.0
0.00
0
0.00
Table 3: Comparative results on the FCT and EOL datasets across training and validation cohorts. β is the instability sensitivity parameter, and R^(π) denotes empirical escape risk.
Figure 4: Cost-risk Pareto fronts for the two test suites. The grey dotted line shows a logarithmic fit to the Pareto-optimal points, and the red star marks the zero-escape operating point.
Figure 5: Rolling Cred selection fraction during training for the Thompson Sampling agent on the FCT and EOL datasets.
Figure 6: Rolling pass rate and Cred selection behavior during validation for the FCT and EOL datasets.
Reliability qualification of advanced semiconductor devices requires sequential stress decisions that balance characterization objectives against multiple competing failure mechanisms. Current practice relies on static test plans derived from population-level acceleration models, which cannot adapt to per-unit variability or real-time degradation observations. This paper presents a closed-loop adaptive test planning framework that formulates reliability qualification as a partially observable sequential decision problem and solves it using Monte Carlo tree search for seed-action simulators (MCTS-SA) coupled with extended Kalman filter (EKF) belief-state estimation. The framework models stochastic, per-device variability in bias temperature instability (BTI), electromigration (EM), and time-dependent dielectric breakdown (TDDB), and treats stress selection as a constrained sequential optimization, i.e., to maximize the probability of successful degradation characterization while respecting catastrophic failure constraints. Under the experimental assumptions used here (discrete stress actions, proxy damage observability, and cumulative degradation without recovery), we believe this to be a novel application of tree-search-based adaptive test planning to multi-mechanism reliability qualification. Across 5,000 planning iterations, the characterization yield (CY) improves from 20% in the first 500 iterations to over 54% in the final 500, with 39% cumulative success, while the best successful test sequence terminates with EM and TDDB damage fractions DEM=0.564 and DTDDB=0.537, well within safety margins. These results demonstrate that sequential Bayesian planning can synthesize damage-aware test policies that significantly outperform non-adaptive strategies for reliability qualification under competing failure modes.
Youssef A. Elhagrasy, Ian Hill, André Ivanov
Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
This paper presents a probabilistic framework for online test-time adaptation problems. In them, a model is trained on labeled data but must adapt to unlabeled data at test time under the assumption that training and test distributions potentially differ, that is, there might have been a distributional shift. The framework is based on a state-space modelling architecture from which parameter learning, parameter time evolution, prior tuning, and prediction can be characterized.
Daniel Corrales, David Ríos Insua
Escuela de Doctorado, Universidad Aut´onoma de Madrid, 28049 Madrid, Spain · Institute of Mathematical Sciences, ICMAT-CSIC, 28049 Madrid, Spain
Objective: We investigate whether accounting for epistemic uncertainty can improve the reliability of automated defect detection in medical device manufacturing. Methods: We consider a machine learning framework that operates on heterogeneous manufacturing and device-report data represented with Knowledge Graphs. To mitigate errors arising from uncertainty in the decision model, we analyze a principled rejection strategy to abstain from predictions whose estimated epistemic uncertainty exceeds a specified threshold. We evaluate the approach using standard synthetic benchmarks and real-world medical device report data. Results: The theoretical results establish the validity of the method characterizing the regimes under which it is expected to be effective. Empirically, the rejection strategy enables explicit control of coverage, that is, the proportion of samples for which the model issues predictions, while improving performance on the retained samples. On 266,170 real-world FDA MAUDE device reports, a 10% abstention rate reduces classification error by 48%, and more aggressive rejection (approximately 70% coverage) yields near-perfect accuracy on the retained samples. On standard synthetic manufacturing benchmarks, abstaining on 9% of the decisions, our approach reduces the risk up to 63% compared with the standard no-abstention approach. Conclusions: Abstaining from predictions with high epistemic uncertainty can provide a practical tool for controlling the reliability of machine learning-based defect detection, especially in high-stakes medical device manufacturing applications. Significance: Uncertainty-aware defect detection may support safer and more reliable quality assurance in medical device manufacturing by identifying cases that require additional inspection rather than issuing potentially harmful predictions.
Raham A. Butt, Marco Romanelli, Roche C. de Guzman
Department of Engineering, Hofstra University, Hempstead, NY 11549 USA · Department of Computer Science, Hofstra University, Hempstead, NY 11549 USA