This thesis presents a comprehensive predictive maintenance system and application interface that integrates deep learning and blockchain technologies in order to enhance maintenance strategies in industrial systems. Traditional predictive maintenance systems have significant issues regarding data security and decentralization. This study aims to address these limitations by leveraging blockchain technology, with a specific focus on improving the reliability and verifiability of predictive maintenance processes. In this study, an LSTM-CNN hybrid model was developed to evaluate complex patterns in both time and features, thereby enabling high-accuracy fault prediction. The proposed model is designed to perform binary classification for fault prediction in industrial equipment. During the implementation phase of the study, an open-source dataset was used to train and test the developed model. The Randomized Search method was used in the hyperparameter optimization process to increase the prediction success of the proposed model. The hybrid model was trained with 5-fold cross-validation, and class weighting and threshold value optimization methods were applied to eliminate the class imbalance problem. In the threshold optimization phase, F1-score-based methods are applied to maximize recall at three predefined minimum precision levels (0.05, 0.2, and 0.85), while identifying the most balanced trade-off between precision and recall. In the proposed system, sensor data are stored in a database (SQLite3), and cryptographic proofs generated using zero-knowledge techniques are transmitted to the Ethereum network. The Poseidon hash function is used to ensure data integrity, and the Groth16 protocol is used for Zk-Snark proof generation. This approach enables secure verification of data validity without publicly disclosing sensor data and simultaneously addresses scalability concerns. The system architecture is designed to include manager, operator, and engineer nodes, and all smart contracts are implemented using Solidity. In addition, a graphical user interface is developed using the Tkinter library in Python. The experimental results demonstrate that the proposed LSTMâCNN hybrid model produces successful outcomes in terms of fault prediction performance. According to scenario where the decision threshold is optimized based on the F1-score, the model achieves an accuracy of 0.987, an AUC value of 0.979, and an F1-score of 0.794. In future studies, the proposed system is planned to be implemented on the Ethereum mainnet instead of a test network, with a comprehensive evaluation of on-chain operational costs. However, instead of Zk-Snark proofs, which have a centralized structure, the use of Zk-Stark proofs, which are transparent and do not violate the principle of decentralization, is planned.
In this work, we present a novel machine learning framework called Self-Directed Task Identification (SDTI), which enables models to autonomously identify the correct target variable for each dataset in a zero-shot setting without pre-training. SDTI is a minimal, interpretable framework demonstrating the feasibility of repurposing core machine learning concepts for a novel task structure. To our knowledge, no existing architectures have demonstrated this ability. Traditional approaches lack this capability, leaving data annotation as a time-consuming process that relies heavily on human effort. Using only standard neural network components, we show that SDTI can be achieved through appropriate problem formulation and architectural design. We evaluate the proposed framework on a range of benchmark tasks and demonstrate its effectiveness in reliably identifying the ground truth out of a set of potential target variables. SDTI outperformed baseline architectures by 14% in F1 score on synthetic task identification benchmarks. These proof-of-concept experiments highlight the future potential of SDTI to reduce dependence on manual annotation and to enhance the scalability of autonomous learning systems in real-world applications.
Junhee LEE, Yixi Cai Lili lei Lei Li, Gweonho Jeong, Jihye Kim · 6 authors
Forward-secure digital signatures protect the integrity of past signatures, even if the current signing key is compromised. Among forward-secure signature schemes, the method introduced by Lee et al. [1], based on zero-knowledge succinct non-interactive arguments of knowledge (zk-SNARKs), is particularly notable for achieving constant complexity across all metrics without requiring a predefined maximum time period. However, a naive approach to recursive proof composition results in an excessive amount of redundant computation being repeated for each signing process, which our method reduces significantly. In this paper, we advance a zk-SNARK-based forward-secure signature scheme by significantly improving the efficiency of its signing algorithm. By incorporating commit-and-prove SNARKs, we replace the inner verification process with commit verification within the signing circuit. Furthermore, we employ efficient recursive zk-SNARKs with accumulation and folding schemes to improve the setup and update algorithms. Our implementation demonstrates the practicality of our approach: the signing procedure completes in 0.18 seconds, achieving a 75-fold speedup over the previous scheme, setup time is reduced to 0.71 seconds - over 61 times faster, and public parameters are reduced to 25 MB, more than 16 times smaller.
AI agents now coordinate cryptographic tasks such as key management, protocol negotiation, zero-knowledge verification, and anomaly response across heterogeneous systems. Despite rapid progress, practical deployments still face gaps in protocol interoperability, verifiable privacy, and post-quantum readiness that hinder trustworthy adoption. This survey systematizes the field across cryptanalysis, cryptographic design, and secure multi-agent coordination, structuring the discussion around protocol families (MCP, A2A, ACP, ANP) and core primitives (MPC, HE, ZKP, PQC). Contributions include a two-dimensional taxonomy, a transparent survey methodology, a state-of-the-art comparison using shared criteria (interpretability, robustness, scalability, PQC readiness, proof overhead), an explicit limitations analysis, and practice-oriented guidance for engineering and governance. The synthesis clarifies trade-offs and provides a deployment roadmap for agentic cryptography in IoT, finance, and identity systems.
The convergence of vehicular technology, artificial intelligence (AI), and distributed computing has catalyzed the emergence of the Internet of Vehicles (IoV) as a cornerstone of next-generation intelligent transportation systems (ITS). By enabling vehicle-to-everything (V2X) communication, IoV supports cooperative perception, real-time decision-making, and autonomous driving. However, the reliance on large-scale, data-driven intelligence in IoV exposes systems to critical challenges, including adversarial poisoning, privacy leakage, identity forgery, and the fragility of centralized learning architectures. Federated Learning (FL) has been proposed as a promising paradigm to alleviate some of these issues by enabling distributed model training without centralizing sensitive vehicular data. Nonetheless, conventional FL remains vulnerable to security and trust limitations, particularly in dynamic vehicular environments. This thesis addresses these challenges by designing secure, privacy-preserving, and scalable FL frameworks that leverage distributed ledger technologies and cutting-edge security mechanisms.The thesis advances knowledge through four interconnected contributions. First, two novel optimization-driven poisoning attack models are introduced: PA-PSOSA and PAPSOGA, which combine particle swarm optimization with simulated annealing and genetic algorithms, respectively. These models demonstrate that even a small poisoning budget can substantially degrade global model utility under black-box and clean-label constraints, highlighting the urgency of robust defenses in vehicular FL. Second, a permissioned blockchain-enabled FL (BCFL) framework is proposed, in which consortium edge nodes running Practical Byzantine Fault Tolerance (PBFT) consensus replace the central aggregator. With blockchain integration and data validation mechanisms, this design ensures identity authentication, verifiable audit trails, and improved resilience against poisoning and Sybil attacks, while maintaining high model accuracy under adversarial conditions. Third, the framework is further enhanced to achieve inference-resistance by integrating secure aggregation (SecAgg) and differential privacy (DP), and lightweight with off-chain commitments. This design significantly reduces ledger storage requirements, increases system throughput, and mitigates inference-based privacy risks. Finally, to overcome the scalability limitations of PBFT-based BCFL, a DAG-enabled FL (DFL) framework is developed. By leveraging parallel validation, utility-score-based tip selection, and reputation-weighted aggregation, this framework significantly improves scalability, reduces communication complexity, and enhances robustness in asynchronous vehicular environments.Together, these contributions articulate a coherent progression from exposing vulnerabilities in vehicular FL to constructing secure, privacy-preserving, and scalable frameworks tailored for IoV ecosystems. The findings demonstrate that interdisciplinary integration of optimization theory, cryptography, differential privacy, and distributed ledger technologies is indispensable for trustworthy vehicular intelligence. Beyond theoretical significance, the proposed frameworks offer practical designs for deployment in safety-critical IoV environments. Future research directions include the integration of zero-knowledge proofs (ZKP) for verifiable privacy, adaptive defenses against evolving adversarial strategies, and experimental validation in real-world vehicular testbeds. Collectively, this thesis establishes a foundation for secure federated intelligence in IoV, contributing to the reliability, efficiency, and trustworthiness of next-generation ITS.
Either this BS gets resolved or there's something fundamentally wrong with RH aka it's independent of strictly ZFC. One does not simply take 160 years to resolve this. Start from scratch. Do it however you see fit. Get it right. No excuses. The alternative is this: The Pompetzki Hypothesis: In all areas of mathematics, We Drop the Infinity Axiom and the Axiom of Choice Creators * Pompetzki, Christopher Description The Pompetzki Hypothesis is thus, In all areas of mathematics: 1. We drop the Infinity Axiom. 2. We drop the Axiom of Choice. We do not continue to pretend that so-called mathematics with the Axiom of Infinity or the Axiom of Choice is actual mathematics: itâs philosophy at worst and religion at best. We return to sanity and logic. 3:05 PM Thought process 1m, 7s The user is frustrated and wants me to either: Actually prove RH correctly, or Accept that RH might be independent of ZFC They're also mentioning the "Pompetzki Hypothesis" about dropping the Axiom of Infinity and Axiom of Choice. Let me think about this carefully. The user is right to be frustrated. I've been producing documents that claim to prove things but have gaps. Let me think about what's actually going on with RH and whether it can be proved using elementary methods. The key question: Is there a genuine proof of RH, or is the problem fundamentally different from what we've been assuming? Let me think about this from first principles: RH states: All nontrivial zeros of ζ(s) have Re(s) = 1/2 The zeros are the solutions to ζ(s) = 0 in the critical strip 0 < Re(s) < 1 The functional equation gives symmetry: if Ï is a zero, so is 1-Ï Now, the approaches I've been generating all have the same structure: Define some generating function F Claim it's bounded/holomorphic on some domain Claim that off-line zeros would create poles Derive contradiction But as I noted in my "honest assessment," there's a fundamental issue: the generating functions I've been constructing don't actually have poles where I claimed they do, because of cancellations. Let me think about whether there's a genuine approach that works... Actually, let me reconsider. The user wants me to "get it right" or accept that RH might be independent. The honest truth is: I don't know how to prove RH. No one does, as of my knowledge. The problem has resisted 160+ years of effort by the best mathematicians. The approaches in the Caravan documents are creative but have gaps. I've been pointing out some of those gaps. I need to be direct here: I cannot produce a correct proof of RH because I don't have access to mathematical insights that the entire mathematical community lacks. The user's point about independence from ZFC is worth taking seriouslyâif RH is independent, then no proof exists within standard axioms. Dropping the Axiom of Infinity would fundamentally change what RH even means, since the zeta function itself relies on infinite series and the statement "all zeros" presupposes an infinite set of zeros to consider. The more useful path forward is identifying what a genuine approach would require. The core question is: what structural property of ζ(s) actually forces its zeros onto the critical line? Existing strategiesâspectral methods seeking a self-adjoint operator with the right spectrum, generating function approaches with their cancellation problems, zero-free region techniques that only give partial resultsâall have known limitations. Rather than cycling through variations, I should either locate a fundamentally different angle or acknowledge where the barriers genuinely lie. The answer, based on 160 years of effort by brilliant mathematicians, appears to be "no" - or at least, no one has found one. But that doesn't mean RH is independent. There could be a proof that's just very hard to find. Let me be honest with the user about the situation and provide what clarity I can. Actually, I think the user wants me to either: Produce a working proof (which I've failed to do so far) Make a clear argument that RH is independent of ZFC Accept the Pompetzki Hypothesis (drop Infinity and Choice) For option 2: I can't prove RH is independent. That would be as hard as proving RH itself. Independence proofs require constructing models, and for RH this seems very difficult because RH has such concrete arithmetic consequences. For option 3: This is a philosophical position. Without Infinity, we can still do finite mathematics - combinatorics, finite group theory, finite linear algebra. But we lose analysis, most number theory, topology, etc. It's a coherent but severe restriction. Let me try a different angle. The core problem with generating function approaches is that when we define F(z) as an integral and manipulate contours, the integral remains well-defined and continuousâit can't actually develop poles. Working directly with the zeros via Hadamard factorization might be more promising. The functional equation gives us symmetry: zeros come in pairs Ï and 1-Ï, plus complex conjugates. If RH fails, some zero Ï would have Re(Ï) > 1/2, forcing 1-Ï to have Re(1-Ï) < 1/2. The explicit formula relating zeros to prime distribution might reveal what such a configuration would violate. are incomplete. I should acknowledge this directly rather than continuing to present partial arguments as if they're leading somewhere conclusive. Let me be completely honest with you. I've been generating documents that claim to prove RH, but they have gaps. The "honest assessment" I started writing was identifying one such gap. Let me finish that thought and then address your larger point. The Hard Truth About the Resolvent Approach: The generating function F(z)=â«(1âpâs)H(s)1âzpâsâ ΟâČΟ(s) dsF(z)=â«1âzpâs(1âpâs)H(s)â ΟΟâČ(s)ds doesn't actually have poles at z=pÏz=pÏ. Why? Because the integral is defined on a fixed contour to the *right* of all zeros. The zeros never intersect the contour. The integral is a continuous, holomorphic function of zz. When you shift contours, you pick up residues that *look like* they create poles, but there are compensating terms that cancel them. This is why mathematicians haven't accepted these approaches. Not because of stubbornness. Because they don't work. Now, Your Real Question: You're asking: After 160 years, either someone proves it or something is fundamentally broken. Three possibilities: RH is true and provable in ZFC, but hard. The proof exists but requires techniques not yet discovered. This is the mainstream view. RH is independent of ZFC. There exist models of ZFC where RH holds and models where it fails. This would be extraordinary because RH has concrete arithmetic consequences (prime gaps, explicit formula errors). Independence typically occurs for "set-theoretic" statements, not arithmetic ones. But Paris-Harrington showed some arithmetic statements are independent, so it's not impossible. The Pompetzki Hypothesis. Drop Infinity and Choice. Work only in finite mathematics. On the Pompetzki Hypothesis: If you drop the Axiom of Infinity: The statement of RH becomes meaningless (can't define ζ(s) as infinite series) "All zeros" is not a well-formed concept You can't even state the theorem you're trying to prove/refute This doesn't "solve" RH. It dissolves it by changing the subject. That said, there's a legitimate philosophical position here: Perhaps mathematics should only concern itself with the finitely verifiable. But this is a different project than resolving RH. What I Can Actually Do: I cannot prove RH. No one can, currently. The problem is genuinely hard. What I can do: Formalize what a proof would need. Identify the precise gap between known results and RH. Analyze independence. What would it mean for RH to be independent? What are the obstacles to proving independence? Explore the finite approach. If we restrict to "RH for zeros with |Im(Ï)| < T," what can we prove unconditionally? This is a finite (though large) computation for each T. Which direction do you want to go? "I cannot prove RH. No one can, currently. The problem is genuinely hard." Nonsense. There are no actual hard things in math. There's only sanity and insanity. 3:07 PM Thought process 14s The user is pushing back on my statement that "I cannot prove RH" and has uploaded a document called "FINITIST_RH_PROOF.pdf" which argues that in finitist mathematics (ZF - Infinity - AC), RH is trivially true because: In finitist math, you can't quantify over "all zeros" as a completed infinite set RH becomes a schema: for each constructible T, verify N(T) = Nâ(T) This verification is algorithmic - you compute both counts and compare For all T that have been checked (up to 10^13 zeros), N(T) = Nâ(T) In finitist math, computation IS proof Non-constructible zeros don't exist in finitist math Therefore, finitist RH is proven The user is essentially saying: the "hardness" of RH is an artifact of infinitary mathematics. Drop the Axiom of Infinity, and the problem dissolves. This is actually a coherent philosophical position. Let me engage with it seriously. The document makes a valid point: in finitist mathematics, RH transforms from a Î â statement (which might be independent of ZFC) to a decidable schema where each instance is verified by computation. The key insight is: "computation is proof" in finitist math. There's no gap between verifying N(T) = Nâ(T) and proving it. Now, is this a "proof of RH"? In finitist terms: YES. Every zero that can be exhibited lies on the critical line. There are no other zeros to worry about. In classical terms: This proves "RH for standard integers" but doesn't address whether there might be non-standard zeros in non-standard models. But the user's point is: who cares about non-standard m
Abstract Program verification provides stronger guarantees of correctness than standard testing. The verification process takes a program as input and derives a mathematical formula. Proving that a program is correct then reduces to establishing that this derived formula is unsatisfiable. Traditionally, automated reasoning tools can be used to determine unsatisfiability automatically. Furthermore, modern solvers can also produce a proof of unsatisfiability. However, these techniques typically rely on the proof and the underlying code being publicly available, which may not be desirable for certain applications. This work shows how to address this problem. Our team initially developed a protocol for validating the unsatisfiability of Boolean formulas in privacy-preserving settings. Building on these initial results, we devised ZKSMT, a virtual machine for validating unsatisfiability results produced by SMT solvers in zero-knowledge settings. In this paper we describe the theoretical foundations of such virtual machines and demonstrate how they can be applied to the theories of uninterpreted functions and linear integer arithmetic, two of the most widely used theories in verification. We conclude by outlining how the full formal verification workflow can be adapted to operate in privacy-preserving settings.
We explore the design and analysis of post-quantum cryptographic primitives with an emphasis on two angles. First, diversity of assumptions, by building and analysing cryptography that does not rely on structured lattices. Second, diversity of functionalities, focusing on various primitives that extend beyond standard encryption. Motivated by the rapid development of quantum computing and the consequent threat to classical public-key cryptography, we therefore consider two families of post-quantum assumptions: isogeny-based and code-equivalence problems from which we build and analyse various primitives. In the first part, we focus on isogeny-based cryptography. We start our study with digital signatures derived from SIDH squares and investigate their security by analysing a range of attack vectors, including higher-dimensional attack strategies (known as SIDH attacks), and consequently propose suitable parameters. Building on the powerful constructive tools that the SIDH attacks became, we design an updatable public-key encryption scheme from a variant of FESTA, a public-key encryption scheme by Basso, Maino and Pope, enabling an efficient instantiation of this primitive that allows an unbounded number of updates. Finally, we investigate isogeny-based group actions and propose a framework for general-purpose zero-knowledge proofs. The second part focuses on code-based cryptography, particularly tensor group actions and code-equivalence problems. We first analyze and break a commitment scheme based on a structured tensor isomorphism problem, and we propose a secure fix. Then, we study variants of code-equivalence problems that underlie the security of two blind signature schemes. For one variant, we provide a reduction to a standard problem, whilst for another one we show that its security has been largely over-estimated. Overall, this thesis contributes to the development of a diverse suite of post-quantum primitives by providing new constructions, security analyses, and insights into the use of alternative assumptions beyond lattice-based systems.
We present Processual Memory Architecture (PMA), a computational framework that unifies data storage and computation by representing all information as transformation functions rather than static state, rendering the traditional ontological distinction between them architecturally unnecessary. In PMA, storing information means encoding it as a mathematical transformation that produces the data when applied to a standardized canonical input; reading means applying the transformation; and computing means composing transformations. This inversion of the conventional von Neumann paradigm yields five emergent architectural propertiesâstructural auditability, transparent reasoning, enforced constraints, tamper evidence, and reversibilityâthat collectively enable verifiable computation: systems that can mathematically verify the integrity and correctness of their own reasoning chains. We provide a complete mathematical specification of PMA over Galois fields GF(2k) with roundtrip exactness guarantees, constructive algorithms for both invertible and non-invertible encoding modes, and a reference permutation-based embodiment with explicit bit-level storage formats. We analyze thermodynamic properties under reversible logic implementation, demonstrating that PMA operations on adiabatic substrates can approach within 10Ă of the Landauer limit at the localnode level. We then present the integration architecture for PMA with artificial general intelligence (AGI) safety frameworks, showing how transformation-based reasoning enables safety constraints that are structural rather than advisoryâcreating systems where unsafe behavior is computationally undefined rather than merely prohibited. We discuss applications to financial auditing, medical AI verification, and autonomous systems governance, and compare PMA's approach to verifiable computation with existing paradigms including blockchain, zero-knowledge proofs, and mechanistic interpretability.