arXiv subjects

Jaewon Jang

Publications and source records attributed to Jaewon Jang.

10 recordsLinked to original sources

arXiv2026

SePArate: Segmenting Patterns from Defects in Wafer Manufacturing Using Weak Supervision

In semiconductor manufacturing, defect analysis is essential, but manual inspection cannot scale. However, existing automated inspection methods remain insufficient for root-cause analysis and process optimization. To this end, we present SePArate, a weakly supervised wafer defect segmentation method. SePArate enables pixel-level separation of patterns by leveraging only image-level annotations. It consists of a three-phase training: encoder pretraining, knowledge transfer to learn spatial cues, and training on synthetic mixed-defect data for accurate segmentation. Experiments demonstrate that SePArate outperforms the baselines.

cs.CV↗

arXiv2026

The Mask Is Not the Model: Auditing Prefix Invariance in Attention, State-Space, and Hybrid Sequence Models

Hybrid sequence models must satisfy prefix invariance: representations at position t must not depend on future inputs, yet this is rarely verified. We formalize prefix invariance and give a lightweight audit, two forward passes, no training or gradients, yielding a per-layer score localizing where causality breaks. Attention-mask inspection, the field's default check, is incomplete: causality is a graph-level property, and leaks can occur via scans, aggregations, or normalization despite correct masks. Across 192 injected-fault trials on eight checkpoints, mask inspection detected none, while our audit localized all 192/192 to the exact layer. Static/dynamic analysis of chunked-scan code in transformers found the same defect in Zamba2 and Nemotron-H, an inter-chunk axis error fixed via the reference implementation. The method fits on one page and runs in seconds.

cs.LG↗

arXiv2026

A Numerical Realization of Suzuki's Weil-Quadratic-Form Operator: The Archimedean Spectral Law, its Universality, and an Operator Form of Weil's Positivity Criterion

This paper presents the first numerical realization of Suzuki's Weil-Quadratic-Form operator, a candidate for the Hilbert--Pólya program linking spectral positivity to the Riemann Hypothesis (RH). Suzuki's 2026 construction was purely theoretical; here, the operator is instantiated via P1 finite-element discretization and Richardson extrapolation. Key results include: (R1) In the prime-free regime, the spectrum follows a closed Archimedean law $A_k(a) = \log(1/a) + \log(k-2) + B_0 + O(a)$, with $B_0 = \log q - 2\log 2$, confirmed to 30-digit precision. (R2) A Mellin double-pole argument proves the head coefficient $B(ν)$ and shows $B_0$ depends only on the conductor $q$, independent of the Archimedean parameter. (R2b) The degree $d$ of an L-function appears directly as the logarithmic slope of the spectrum. (R3) Total spectral intensity follows the prime number theorem, $S(a) \sim (2a)^3/6$. (R4) Nontrivial zeros are not eigenvalues but occur in the explicit-formula error term of the prime symbol. (R5) The best-match line $σ^*(a)$ descends toward the critical line. (R6) Weil's positivity criterion is realized in operator form: bounded residual growth corresponds to all zeros on the line, while an injected off-line zero causes exponential blow-up. (R7) The lowest eigenvalue $λ_1(a)$ is strictly positive, decays superexponentially, and passes smoothly through the first prime threshold. (R8) The characteristic function $W(a,0;z)$ is computed for the first time, with all zeros confirmed real. (R9) Indirect traces of GUE statistics appear in the moment structure, even where direct detection is blocked. The authors emphasize that this work does not prove RH. All results are Archimedean and universal, with significance lying in the faithful numerical realization of classical identities rather than new arithmetic.

math.GM↗

arXiv2026

Placement Is Free, Composition Is Not: The Latin Square as a Provably-Balanced Construction for Heterogeneous Sequence-Mixer Stacks

Since GPT, most Transformers have repeated the same attention mechanism at every layer. Yet this design is largely a convention rather than a tested conclusion. When multiple sequence mixers are combined in one stack, improvements may arise from mechanism choice, placement, or both, making causal attribution difficult. We introduce Aether-7B-5Attn, a 6.59B-parameter mixture-of-experts model ($\approx$2.98B active) whose 49 layers contain seven sequence-mixing mechanisms arranged as a $7\times7$ Latin square. Because each mechanism appears exactly once in every row and column, the design guarantees balanced exposure across depth while eliminating placement confounds. To evaluate this principle, we build a parameter-matched proxy with four mechanisms arranged as a $4\times4$ Latin square over sixteen layers, matched to 700.9M parameters and trained with eight seeds per arm. The results reveal a clear dissociation. Rearranging a distributed heterogeneous stack into a balanced periodic cycle changes validation loss by only 0.16\%, indicating that exact placement has little effect. In contrast, clustering the same mechanisms into contiguous depth bands incurs a 0.59\% penalty, while replacing the heterogeneous stack with a homogeneous one incurs a 1.68\% penalty. These results indicate that performance depends primarily on heterogeneous composition distributed across depth rather than on any particular permutation. We confirm this finding at 2.16$\times$ larger scale (1.514B parameters), where the homogeneous-stack penalty increases to 2.63\% and removing the SSM-family mechanism produces a 3.20\% degradation. We further report per-mechanism cost profiles, English and Korean evaluations, and a causal-safety audit of all 49 layers. We release model weights, training recipes, training code, logs, and architecture source code.

cs.LG↗

arXiv2026

Quantum Cryptanalysis on IBM Quantum Hardware: Extending Even--Mansour Period Recovery from $N=4$ to $N=10$

We report genuine-un-compiled, textbook-faithful-quantum cryptanalysis of symmetric-cipher structures executed on real IBM quantum hardware (ibm\_kingston, Heron generation). Using Simon's algorithm we recover the hidden period of the Even-Mansour cipher up to security parameter N = 10 on real hardware, beyond the largest previously reported real-hardware key recovery of N = 4, and we cleanly recover the periods of a 3-round Feistel (DES-family) construction at block sizes 6 and 8; a 21-qubit block-10 instance is verified in simulation and submitted to hardware. We further provide a breadth-first benchmark of five genuine quantum attacks spanning four symmetric-cipher design paradigms -- Bernstein-Vazirani (linear structure, single query), Grover (SPN key search, quadratic), and Simon (Even-Mansour, CBC-MAC forgery, and Feistel; exponential-to-polynomial in query complexity) -- validated to the classical-simulation ceiling of 25 qubits. We are deliberately explicit about scope: these attacks target reduced or structured constructions in the Q2 (quantum-query) model, asymptotically follow the birthday bound and therefore do not constitute quantum advantage over classical collision-finding, do not break full AES/RSA or 16-round DES, and rely on error mitigation rather than fault-tolerant error correction. Our contribution is the real-hardware demonstration at record structure sizes, the breadth of genuine algorithmic coverage across four paradigms, and an honest, reproducible benchmark with public artifacts.

cs.CR↗

arXiv2026

Darwin Family: MRI-Trust-Weighted Evolutionary Merging for Training-Free Scaling of Language-Model Reasoning

We present Darwin Family, a framework for training-free evolutionary merging of large language models via gradient-free weight-space recombination. We ask whether frontier-level reasoning performance can be improved without additional training, by reorganizing latent capabilities already encoded in existing checkpoints. Darwin introduces three key ideas: (i) a 14-dimensional adaptive merge genome enabling fine-grained component- and block-level recombination; (ii) MRI-Trust Fusion, which adaptively balances diagnostic layer-importance signals with evolutionary search through a learnable trust parameter; and (iii) an Architecture Mapper that enables cross-architecture breeding between heterogeneous model families. Empirically, the flagship Darwin-27B-Opus achieves 86.9% on GPQA Diamond, ranking #6 among 1,252 evaluated models, and outperforming its fully trained foundation model without any gradient-based training. Across scales from 4B to 35B parameters, Darwin models consistently improve over their parents, support recursive multi-generation evolution, and enable a training-free evolutionary merge that combines Transformer- and Mamba-based components. Together, the Darwin Family demonstrates that diagnostic-guided evolutionary merging is a practical and reproducible alternative to costly post-training pipelines for reasoning-centric language models.

cs.NE↗

arXiv2025

Can Consumer Chatbots Reason? A Student-Led Field Experiment Embedded in an "AI-for-All" Undergraduate Course

Claims about whether large language model (LLM) chatbots "reason" are typically debated using curated benchmarks and laboratory-style evaluation protocols. This paper offers a complementary perspective: a student-led field experiment embedded as a midterm project in UNIV 182 (AI4All) at George Mason University, a Mason Core course designed for undergraduates across disciplines with no expected prior STEM exposure. Student teams designed their own reasoning tasks, ran them on widely used consumer chatbots representative of current capabilities, and evaluated both (i) answer correctness and (ii) the validity of the chatbot's stated reasoning (for example, cases where an answer is correct but the explanation is not, or vice versa). Across eight teams that reported standardized scores, students contributed 80 original reasoning prompts spanning six categories: pattern completion, transformation rules, spatial/visual reasoning, quantitative reasoning, relational/logic reasoning, and analogical reasoning. These prompts yielded 320 model responses plus follow-up explanations. Aggregating team-level results, OpenAI GPT-5 and Claude 4.5 achieved the highest mean answer accuracy (86.2% and 83.8%), followed by Grok 4 (82.5%) and Perplexity (73.1%); explanation validity showed a similar ordering (81.2%, 80.0%, 77.5%, 66.2%). Qualitatively, teams converged on a consistent error signature: strong performance on short, structured math and pattern items but reduced reliability on spatial/visual reasoning and multi-step transformations, with frequent "sound right but reason wrong" explanations. The assignment's primary contribution is pedagogical: it operationalizes AI literacy as experimental practice (prompt design, measurement, rater disagreement, and interpretability/grounding) while producing a reusable, student-generated corpus of reasoning probes grounded in authentic end-user interaction.

cs.CY↗

arXiv2025

Bootstrapping Conditional Retrieval for User-to-Item Recommendations

User-to-item retrieval has been an active research area in recommendation system, and two tower models are widely adopted due to model simplicity and serving efficiency. In this work, we focus on a variant called \textit{conditional retrieval}, where we expect retrieved items to be relevant to a condition (e.g. topic). We propose a method that uses the same training data as standard two tower models but incorporates item-side information as conditions in query. This allows us to bootstrap new conditional retrieval use cases and encourages feature interactions between user and condition. Experiments show that our method can retrieve highly relevant items and outperforms standard two tower models with filters on engagement metrics. The proposed model is deployed to power a topic-based notification feed at Pinterest and led to +0.26\% weekly active users.

cs.IR↗

arXiv2024

Personalized Federated Learning via Sequential Layer Expansion in Representation Learning

Federated learning ensures the privacy of clients by conducting distributed training on individual client devices and sharing only the model weights with a central server. However, in real-world scenarios, the heterogeneity of data among clients necessitates appropriate personalization methods. In this paper, we aim to address this heterogeneity using a form of parameter decoupling known as representation learning. Representation learning divides deep learning models into 'base' and 'head' components. The base component, capturing common features across all clients, is shared with the server, while the head component, capturing unique features specific to individual clients, remains local. We propose a new representation learning-based approach that suggests decoupling the entire deep learning model into more densely divided parts with the application of suitable scheduling methods, which can benefit not only data heterogeneity but also class heterogeneity. In this paper, we compare and analyze two layer scheduling approaches, namely forward (\textit{Vanilla}) and backward (\textit{Anti}), in the context of data and class heterogeneity among clients. Our experimental results show that the proposed algorithm, when compared to existing personalized federated learning algorithms, achieves increased accuracy, especially under challenging conditions, while reducing computation costs.

cs.LG↗

arXiv2020

Application of Genetic Algorithm for More Efficient Multi-Layer Thickness Optimization in Solar Cells

Thin-film solar cells are predominately designed similar to a stacked structure. Optimizing the layer thicknesses in this stack structure is crucial to extract the best efficiency of the solar cell. The commonplace method used in optimization simulations, such as for optimizing the optical spacer layers' thicknesses, is the parameter sweep. Our simulation study shows that the implementation of a meta-heuristic method like the genetic algorithm results in a significantly faster and accurate search method when compared to the brute-force parameter sweep method in both single and multi-layer optimization. While other sweep methods can also outperform the brute-force method, they do not consistently exhibit $100\%$ accuracy in the optimized results like our genetic algorithm. We have used a well-studied P3HT-based structure to test our algorithm. Our best-case scenario was observed to use $60.84\%$ fewer simulations than the brute-force method.

cs.NE↗