Zero-Knowledge Proofs (ZKP) are protocols which construct cryptographic proofs to demonstrate knowledge of a secret input in a computation without revealing any information about the secret. ZKPs enable novel applications in private and verifiable computing such as anonymized cryptocurrencies and blockchain scaling and have seen adoption in several real-world systems. Prior work has accelerated ZKPs on GPUs by leveraging the inherent parallelism in core computation kernels like Multi-Scalar Multiplication (MSM). However, we find that a systematic characterization of execution bottlenecks in ZKPs, as well as their scalability on modern GPU architectures, is missing in the literature. This paper presents ZKProphet, a comprehensive performance study of Zero-Knowledge Proofs on GPUs. Following massive speedups of MSM, we find that ZKPs are bottlenecked by kernels like Number-Theoretic Transform (NTT), as they account for up to 90% of the proof generation latency on GPUs when paired with optimized MSM implementations. Available NTT implementations under-utilize GPU compute resources and often do not employ architectural features like asynchronous compute and memory operations. We observe that the arithmetic operations underlying ZKPs execute exclusively on the GPU's 32-bit integer pipeline and exhibit limited instruction-level parallelism due to data dependencies. Their performance is thus limited by the available integer compute units. While one way to scale the performance of ZKPs is adding more compute units, we discuss how runtime parameter tuning for optimizations like precomputed inputs and alternative data representations can extract additional speedup. With this work, we provide the ZKP community a roadmap to scale performance on GPUs and construct definitive GPU-accelerated ZKPs for their application requirements and available hardware resources.
Nominative signatures allow us to indicate who can verify a signature, and they can be employed to construct a non-transferable signature verification system that prevents the signature verification by a third party in unexpected situations. For example, this system can prevent IOU/loan certificate verification in unexpected situations. However, nominative signatures themselves do not allow the verifier to check whether the funds will be transferred in the future or have been transferred.It would be desirable to verify the fact simultaneously when the system involves a certain money transfer such as cryptocurrencies/cryptoassets. In this paper, we propose a smart contract-based non-transferable signature verification system using nominative signatures. We pay attention to the fact that the invisibility, which is a security requirement to be held for nominative signatures, allows us to publish nominative signatures on the blockchain. Our system can verify whether a money transfer actually will take place, in addition to indicating who can verify a signature. We transform the Hanaoka-Schuldt nominative signature scheme (ACNS 2011, IEICE Trans. 2016) which is constructed over a symmetric pairing to a scheme constructed over an asymmetric pairing, and evaluate the gas cost when a smart contract runs the verification algorithm of the modified Hanaoka-Schuldt nominative signature scheme.
Identity-based cryptography (IBC), proposed by Adi Shamir, revolutionized public key authentication by eliminating the need for certificates, enabling a more efficient and scalable approach to cryptographic systems. Meanwhile, in \cite{Katsumata2024group}, Katsumata et al. were the first to present the blind signature protocol based on the hardness assumption of isogeny with provable security, which resembles the Schnorr blind signature. Building upon these foundational concepts, we propose an Identity-Based Blind Signature Scheme with an Honest Zero-Knowledge Verifier utilizing the CSIDH framework. This scheme combines blind signatures for privacy preservation with zero-knowledge proofs to ensure the verifier's honesty without revealing any additional information. Leveraging the quantum-resistant properties of CSIDH, a post-quantum secure scheme based on supersingular isogenies, our scheme offers strong protection against quantum adversaries while maintaining computational efficiency. We analyze the security of the introduced protocol in the standard cryptographic model and demonstrate its effectiveness in safeguarding privacy and verifier honesty. Furthermore, we present a performance evaluation, confirming the practical viability of this quantum-resistant cryptographic solution for privacy-preserving applications. This work advances the creation of secure, and scalable cryptographic systems for the post-quantum era.
The advent of decentralized finance has ushered in a transformative era in the financial sector, leveraging blockchain technology to facilitate peer-to-peer transactions without traditional intermediaries. Amidst this innovation, the DeFi landscape faces the pervasive threat of rugpulls, where developers abruptly abandon projects post-fundraising, leaving investors with devalued assets. This growing concern highlights a critical research gap in the proactive detection and prevention of such fraudulent schemes. To combat this, we propose RUGSCREENER, a temporal graph neural network-based solution to identify rugpull risks within DeFi transactions. It employs a dynamic representation of blockchain interactions, enriched with comprehensive node attributes and effective temporal graph learning techniques based on memory and attention mechanisms, effectively capturing the rapid-moving and complex transaction patterns indicative of potential fraud. Our evaluation is based on a newly compiled Ethereum dataset that includes two subsets: an unlabeled set with 1,882,114 transactions from 29,595 tokens for temporal graph representation learning, and a labeled set with 128,819 transactions from 1,000 tokens (500 rugpull and 500 benign) for downstream evaluation. Using this dataset, RUGSCREENER achieves a balanced accuracy of 95.7% in detecting rugpull tokens. Our extensive evaluation, utilizing the Ethereum dataset comprising 1000 tokens, showcases its robust performance with a balanced accuracy of 95.7% in detecting rugpull tokens. Remarkably, RUGSCREENER surpasses existing state-of-the-art graph learning models in detecting rugpull tokens with enhanced accuracy and reliability.
Biometric encryption, especially those that are fingerprint-based, plays a dual, important role in identity authentication as well as in privacy protection. Considering the crucial issue of verifying voters in electronic voting systems, ensuring voter privacy and securing sensitive biometric data is considered important. We provide an innovative approach to the integration of fingerprint authentication through biometrics with zero-knowledge proof, thus serving dual purposes of robustness in identity verification while retaining confidentiality at the level of individual voter data. We have a two-node setup architecture in which one is the Prover Node, and the other is the Verifier Node. It will process the fingerprint of the voter to create a key point cryptographic hash and transmit it to the Verifier Node by using ZKP techniques that assure no revealing of sensitive information. This will be further tested rigorously through performance metrics like latency, bandwidth, CPU usage, and memory consumption. Results have shown very low latency times in relationship to both the Prover Node and the Verifier Node, very low bandwidth usage, and efficient use of CPU and memory, which underlines efficiency and capability related to resource management of the system. The integration of biometric data with ZKP not only enhances the security and privacy of electronic voting systems but also proves viable for real-world application. The findings advocate for the broader adoption of such technologies in digital democracy initiatives, potentially transforming future electoral processes.
Biometric Identification and Security
Advanced Steganography and Watermarking Techniques
Jelle Piepenbrock, Josef Urban, Konstantin Korovin, Miroslav Olšák · 6 authors
The development of strong CDCL-based propositional (SAT) solvers has greatly advanced several areas of automated reasoning (AR). One of the directions in AR is therefore to make use of SAT solvers in expressive formalisms such as first-order logic, for which large corpora of general mathematical problems exist today. This is possible due to Herbrand's theorem, which allows reduction of first-order problems to propositional problems by instantiation. The core challenge is synthesizing the appropriate instances from the typically infinite Herbrand universe. In this work, we develop a machine learning system targeting this task, addressing its combinatorial and invariance properties. In particular, we develop a GNN2RNN architecture based on a graph neural network (GNN) that learns from problems and their solutions independently of many symmetries and symbol names (addressing the abundance of Skolems), combined with a recurrent neural network (RNN) that proposes for each clause its instantiations. The architecture is then combined with an efficient ground solver and, starting with zero knowledge, iteratively trained on a large corpus of mathematical problems. We show that the system is capable of solving many problems by such educated guessing, finding proofs for 32.12% of the training set. The final trained system solves 19.74% of the unseen test data on its own. We also observe that the trained system finds solutions that the iProver and CVC5 systems did not find.
Open access
Natural Language Processing Techniques
Handwritten Text Recognition Techniques
Mathematics, Computing, and Information Processing
Multiple sequence alignment (MSA) is a fundamental algorithm in bioinformatics. In a situation when the alignment might need to be protected while revealing the other information such the input sequences and the alignment score, zero knowledge proof can be used. In this paper, a validator checks the consistency between the input sequence and the alignment, and between the alignment and the alignment score. The validator is written in Circom language which will be compile into a circuit. Using a zero knowledge prove system called zkSNARK, a cryptographic proof is generates for the circuit and its input. This proof demonstrates that all inputs are consistent without revealing the actual alignment.
Himanshu Tiwari, Ayush Raj, Ujjwal Kr. Singh, Hoor Fatima
Incorporating generative artificial intelligence (AI) into design and art has upended established creative paradigms, sparking discussions on the validity of AI-generated art and the development of non-fungible token (NFT) marketplaces. The US Copyright Office rendered a significant decision in February 2023 that highlights the contentious nature of AI work and the need of human intervention in its commercialization. This paper traces the development of artificial intelligence in neural networks and examines how it has affected visual arts. We investigate the idea of autonomously creating digital art in the NFT style utilizing generative adversarial networks (GANs), with striking results. Our work links deep learning and blockchain, enabling AI to find a place in the digital art market.
Generative Adversarial Networks and Image Synthesis
With the rise of "Metaverse" and "Web 3.0", Non-Fungible Token (NFT) has emerged as a kind of pivotal digital asset, garnering significant attention. By the end of March 2024, more than 1.7 billion NFTs have been minted across various blockchain platforms. To effectively locate a desired NFT, conducting searches within a vast array of NFTs is essential. The challenge in NFT retrieval is heightened due to the high degree of similarity among different NFTs, regarding regional and semantic aspects. In this paper, we will introduce a benchmark dataset named "NFT Top1000 Visual-Text Dataset" (NFT1000), containing 7.56 million image-text pairs, and being collected from 1000 most famous PFP1 NFT collections2 by sales volume on the Ethereum blockchain. Based on this dataset and leveraging the CLIP series of pre-trained models as our foundation, we propose the dynamic masking fine-tuning scheme. This innovative approach results in a 7.4\% improvement in the top1 accuracy rate, while utilizing merely 13\% of the total training data (0.79 million vs. 6.1 million). We also propose a robust metric Comprehensive Variance Index (CVI) to assess the similarity and retrieval difficulty of visual-text pairs data. The dataset will be released as an open-source resource. For more details, please refer to: https://github.com/ShuxunoO/NFT-Net.git.
<abstract> <p>This paper proposes a deterministic nonce generation technique to address the catastrophic issues associated with nonce reuse in message signing and to enhance the efficiency of Schnorr multi-signature schemes. Additionally, this research aims to reduce computational complexity and bandwidth requirements in digital and multi-signature schemes while maintaining robust security against common attacks. The proposed method was inspired by the EdDSA approach. The methodology includes a comprehensive mathematical analysis of digital signature algorithms and a rigorous examination of their vulnerabilities to well-known cryptographic attacks. This analysis evaluates the effectiveness and robustness of the proposed nonce generation technique within the frameworks of the Schnorr digital signature and the two-round MuSig schemes. Techniques and tools employed in this research involve deterministically generating nonces by hashing the private key and subsequently hashing the result with the message. Furthermore, it is proposed to exclude the public nonce R from the challenge calculations and to allow signers to directly prove possession of their secret keys through the aggregated public key, thereby eliminating the need for non-interactive zero-knowledge (NIZK) proofs. The findings demonstrate significant reductions in computational complexity and operational requirements, thereby improving bandwidth efficiency and making this method well-suited for resource-constrained devices. The approach also exhibits strong resistance to various attacks, including nonce reuse, key cancellation, rogue keys, and virtual machine rewinding.</p> </abstract>
Open access
Handwritten Text Recognition Techniques
Advanced Steganography and Watermarking Techniques
Abstract: Certificates help students not only to prove their achieved goals and milestones but also ensure that he/she maintains a high level of knowledge in that particular field. An estimated total of 25.57 crore Indian students have been enrolled for primary to higher education in 2020–21 and nearly 65 lakhs of them graduate each year. Throughout this journey, a student generates a myriad number of certificates that may include results, transcripts, degrees, diplomas, etc. A student has to submit these certificates to apply for a job or seek higher admission in any particular organization. A major problem today is manually verifying and authenticating these certificates. Many hardworking people with genuine certificates get rejected and suffer because of the lack of a system that can differentiate original certificates from fake ones. With easy access to cheap and advanced software, document forgery has become a matter of concern nowadays. This scenario demands an updated system that could not only store documents safely but also help verify and authenticate them, their issuers, and holders in a way that is much simpler, effective, and secure. Blockchain technology comes up as a solution to all these problems. Blockchain has recently emerged as a potential means for the document-authentication process and can be easily used to tackle document forgery and counterfeiting as it follows a decentralized approach. Our proposed model includes several methods such as unique hash generation, key cryptography, digital ledgers, proof of work, digital signatures, and distributed storage which has made the document-verification process easier and more secure for both the certificate-generating organization and the holder of the certificate. The SHA-256 algorithm has been used to assign a unique hash to each uploaded document which can be used to validate its authenticity. Thus, this system meets up all the criteria for a document verification system by overcoming the drawbacks and difficulties currently faced in the traditional methods of document verification.
Zero-knowledge Proof (ZKP) allows active image authentication to prove image integrity after editing without revealing its source. However, existing ZKP solutions require impractical execution time, leaving a considerable gap between theory and practice. To tackle this, we present a Region-Aware Photo Assurance System based on the nature of image editing with privacy: sensitive information is usually local and relatively small, and thus by cropping and/or adding mosaic to these small regions suffices to protect privacy. Using ZKP for the locally edited region and digital signature for the others can still ensure the integrity of an image, with significantly better efficiency. We comprehensively analyzed the system's performance and showed the advantage of our system compared with the state-of-the-art ZKP-based method, PhotoProof, with 15x/60xfaster on the KeyGen/Proof operations and 25x lower in the Proof size. Furthermore, we protect several real-world images selected in the Redaction dataset with our system. Our system achieves up to 2,700x faster than PhotoProof for a proof generation. We expect the system can become a practical system for real-world applications.
Digital Media Forensic Detection
Advanced Steganography and Watermarking Techniques
Henri Bouma, Armin Reuter, Patrick Brouwer, Martin George · 19 authors
Authentication of travel documents (e.g., passports) and breeder documents (e.g., birth certificates) is important to facilitate legal movement of passengers and to prevent cross-border crime, such as terrorism, smuggling, illegal migration and human trafficking. However, it is time consuming and difficult to verify all security features, the border guards differ in experience and expertise, and it is hard to stay alert every minute of a working day. New (artificial-intelligence based) technologies can assist in the automated fraud detection in travel and breeder documents, which may lead to faster and more consistent checks. This paper presents five categories of new technologies in automated document authentication to overcome the limitations of current document analysis systems in automated and non-automated border control scenarios. The first category consists of techniques related to the verification of visual security features on the holder page of travel documents. This category includes the verification of KINEGRAMs and other Optically Variable features under different light sources and lighting angles, and the analysis of printing techniques. The second category consists of techniques related to the analysis of breeder documents. This analysis can be at detail level (e.g., investigation of stamps) and at tactical level (e.g., verification of a check digit in a document number). The third category concerns the analysis of travel patterns, using information from the visa pages in passports. The stamps on these pages can be used to extract a travel pattern to support risk assessments and to detect anomalies. The fourth category is an analysis of the border-guard inspection history based upon a distributed ledger and blockchain technology that enables secure storage and prevents undesired manipulations. The last category analyzes the electronic chip of a passport. The software analyses document signer and country signer certificates on the chip to detect vulnerable cryptographic keys and tactical anomalies.
In this paper we propose a novel system for identity verification by amalgamating online signature verification, machine learning, IOT and blockchain to garner their potentials to cope up and to contain this risk of identity theft specifically in the case of online transactions. In this system signals of roll, pitch and yaw values retrieved from MPU6050 sensor (Inertial Measurement Unit) are analysed using Digital Time Wrapping to obtain DTW minimum distance to verify the identity of the user. In case of cryptocurrencies, we propose a system where private key is not stored anywhere but the same unique private key, assigned to the user by Blockchain, is generated every time with the help of method incorporating biometrics and machine learning. The required data will then be sent to blockchain with the help of IOT system to complete the transaction.
In conventional digital signature techniques, secret information, which is utilized for authentication, is disclosed to the verifier. A new digital signature system for image data is proposed. This system can be used to assert the copyright of image data. In this system, a graph generated from an image which must has a signature and an isomorphic graph is concealed in this image. The ZKIP (zero knowledge interactive proof) for the graph isomorphism is applied to assert the copyright of this image. Consequently the secret information is not disclosed during the authentication process.
Digital Image Processing Techniques
Advanced Steganography and Watermarking Techniques