arXiv subjects

Ning Luo

Publications and source records attributed to Ning Luo.

At least 19 recordsLinked to original sources

arXiv2026

Certified but Private: Scalable Zero-Knowledge Proofs for Neural Network Guarantees

With the growing deployment of machine learning models, formal guarantees of the robustness and fairness of these models have become increasingly important in safety-critical and legal-compliance settings. However, model parameters are often commercial secrets that cannot be disclosed to auditors or end users. To this end, we present PANDA, a scalable system that uses zero-knowledge proofs (ZKPs) to prove the robustness and fairness properties of a model without revealing its private parameters. PANDA is built on top of CROWN, an efficient robustness certification framework that is used in many state-of-the-art formal verification tools for neural networks. The core contribution of PANDA is a novel algorithm for proving linear relaxation bounds for non-linear activation layers, yielding simple, lightweight proofs. Remarkably, our system can generate proofs of local robustness for neural networks with more than 2.9M parameters in 5 minutes, and can verify them in 10 seconds. Prior ZKP-based robustness system rely on exponential-time algorithms that cannot scale to nontrivial networks. In contrast, PANDA scales polynomially in the number of neurons in a network, allowing us to support neural networks 4 orders of magnitude larger than previous approaches with significantly reduced prover overhead.

cs.LG↗

arXiv2026

Generative Optimization for Incentivized Advertising with Global Level Constraints

Incentivized advertising allocates monetary or virtual rewards to drive user engagement, where a key challenge is optimizing continuous incentive magnitudes under strict global constraints. This problem is complicated by high-frequency interactions, delayed feedback, and non-Markovian user dynamics such as fatigue, which limit the effectiveness of existing uplift modeling and constrained reinforcement learning approaches. To address these challenges, we propose GOAL, a constraint-aware generative framework that formulates incentive allocation as a conditional sequence generation problem. GOAL directly generates incentive magnitudes conditioned on user histories and system-level global pressure, and integrates a hierarchical causal state encoder to capture both local behavioral dynamics and long-range dependencies. To enable flexible constraint control, we introduce \textbf{S}afe \textbf{C}onstrained \textbf{P}olicy \textbf{O}ptimization (SCPO), which learns a single generative policy that generalizes across a spectrum of ROI constraints without retraining. Experiments on large-scale real-world data and a synthetic fatigue-aware environment show that GOAL improves long-term revenue and user retention while substantially reducing ROI violation rates compared to strong baselines.

cs.LG↗

arXiv2026

BlindMarket: Enabling Verifiable, Confidential, and Traceable IP Core Distribution in Zero-Trust Settings

We present BlindMarket, an end-to-end zero-trust distribution framework for hardware IP cores. BlindMarket allows two parties, the IP user and the IP vendor, to complete an IP trading process with strong guarantees of verifiability and confidentiality before the transaction, and then traceability after. We propose verification heuristics and adapt the cone of influence-based design pruning to overcome the limited scalability common to cryptographic protocols and the hardness of the underlying hardware verification. We systematically evaluate our framework on a diverse set of real-world hardware benchmarks, and the results demonstrate that BlindMarket effectively completes across a diverse set of real-world hardware IP cores, demonstrating successful verification on 12 out of 13 designs and substantial performance improvements enabled by design pruning and control-flow guided heuristics.

cs.CR↗

arXiv2026

Connect the Dots: Knowledge Graph-Guided Crawler Attack on Retrieval-Augmented Generation Systems

Stealing attacks pose a persistent threat to the intellectual property of deployed machine-learning systems. Retrieval-augmented generation (RAG) intensifies this risk by extending the attack surface beyond model weights to knowledge base that often contains IP-bearing assets such as proprietary runbooks, curated domain collections, or licensed documents. Recent work shows that multi-turn questioning can gradually steal corpus content from RAG systems, yet existing attacks are largely heuristic and often plateau early. We address this gap by formulating RAG knowledge-base stealing as an adaptive stochastic coverage problem (ASCP), where each query is a stochastic action and the goal is to maximize the conditional expected marginal gain (CMG) in corpus coverage under a query budget. Bridging ASCP to real-world black-box RAG knowledge-base stealing raises three challenges: CMG is unobservable, the natural-language action space is intractably large, and feasibility constraints require stealthy queries that remain effective under diverse architectures. We introduce RAGCrawler, a knowledge graph-guided attacker that maintains a global attacker-side state to estimate coverage gains, schedule high-value semantic anchors, and generate non-redundant natural queries. Across four corpora and four generators with BGE retriever, RAGCrawler achieves 66.8% average coverage (up to 84.4%) within 1,000 queries, improving coverage by 44.90% relative to the strongest baseline. It also reduces the queries needed to reach 70% coverage by at least 4.03x on average and enables surrogate reconstruction with answer similarity up to 0.699. Our attack is also scalable to retriever switching and newer RAG techniques like query rewriting and multi-query retrieval. These results highlight urgent needs to protect RAG knowledge assets.

cs.CR↗

arXiv2026

Can Deep Research Agents Retrieve and Organize? Evaluating the Synthesis Gap with Expert Taxonomies

Deep Research Agents increasingly automate survey writing, yet existing benchmarks do not jointly test whether they retrieve the papers experts consider essential and organize those papers into paper-grounded taxonomies. We introduce TaxoBench, a benchmark built from 72 highly cited LLM surveys, 3,815 cited papers, and their expert-authored taxonomies. TaxoBench evaluates systems in two settings: Deep Research mode measures end-to-end retrieval and organization from a topic, while Bottom-Up mode provides the expert paper set and isolates organization. We evaluate leaf-level assignments with ARI and V-Measure and hierarchy-level structure with US-TED, US-NTED, and Sem-Path. Across 7 Deep Research Agents and 16 LLM configurations, the best agent retrieves only 20.92% of expert-cited papers, and none of 70 standard Bottom-Up runs reaches the experts' average depth of 4.86. A controlled probe shows that models which match this depth do so by fragmenting the taxonomy, reducing alignment with the expert reference. We further find that raw Sem-Path remains near a no-organization floor even when a newer model generation gains 3.68 pp ARI; after depth matching, humans lead on all 10 matched surveys by 13.27 pp. These results identify retrieval and hierarchical organization as separate bottlenecks and show why hierarchy metrics must be calibrated before they are used to compare models.

cs.CL↗

arXiv2025

Towards Practical Zero-Knowledge Proof for PSPACE

Efficient zero-knowledge proofs (ZKPs) have been restricted to NP statements so far, whereas they exist for all statements in PSPACE. This work presents the first practical zero-knowledge (ZK) protocols for PSPACE-complete statements by enabling ZK proofs of QBF (Quantified Boolean Formula) evaluation. The core idea is to validate quantified resolution proofs (Q-Res) in ZK. We develop an efficient polynomial encoding of Q-Res proofs, enabling proof validation through low-overhead arithmetic checks. We also design a ZK protocol to prove knowledge of a winning strategy related to the QBF, which is often equally important in practice. We implement our protocols and evaluate them on QBFEVAL. The results show that our protocols can verify 72% of QBF evaluations via Q-Res proof and 82% of instances' winning strategies within 100 seconds, for instances where such proofs or strategies can be obtained.

cs.CR↗

arXiv2025

Reward Balancing Revisited: Enhancing Offline Reinforcement Learning for Recommender Systems

Offline reinforcement learning (RL) has emerged as a prevalent and effective methodology for real-world recommender systems, enabling learning policies from historical data and capturing user preferences. In offline RL, reward shaping encounters significant challenges, with past efforts to incorporate prior strategies for uncertainty to improve world models or penalize underexplored state-action pairs. Despite these efforts, a critical gap remains: the simultaneous balancing of intrinsic biases in world models and the diversity of policy recommendations. To address this limitation, we present an innovative offline RL framework termed Reallocated Reward for Recommender Systems (R3S). By integrating inherent model uncertainty to tackle the intrinsic fluctuations in reward predictions, we boost diversity for decision-making to align with a more interactive paradigm, incorporating extra penalizers with decay that deter actions leading to diminished state variety at both local and global scales. The experimental results demonstrate that R3S improves the accuracy of world models and efficiently harmonizes the heterogeneous preferences of the users.

cs.IR↗

arXiv2025

Hierarchical Mask-Enhanced Dual Reconstruction Network for Few-Shot Fine-Grained Image Classification

Few-shot fine-grained image classification (FS-FGIC) is challenging as it requires distinguishing visually similar subclasses with extremely limited labeled examples. Existing methods suffer from critical limitations: metric-based methods lose spatial information and misalign local features, while reconstruction-based methods underuse hierarchical feature information and lack selective focus on discriminative key regions. We propose the Hierarchical Mask-enhanced Dual Reconstruction Network (HMDRN), integrating dual-layer feature reconstruction with mask-enhanced feature processing. HMDRN leverages complementary visual information from different network hierarchies via learnable weights, balancing high-level semantic representations with mid-level structural details. It incorporates a spatial binary mask-enhanced transformer module that selectively enhances discriminative regions while filtering background noise. On three fine-grained datasets, HMDRN consistently outperforms state-of-the-art methods with both Conv-4 and ResNet-12 backbones. Ablation studies validate each component's effectiveness, showing dual-layer reconstruction enhances inter-class discrimination while mask-enhanced transformation reduces intra-class variations.

cs.CV↗

arXiv2025

Learning Monotonic Probabilities with a Generative Cost Model

In many machine learning tasks, it is often necessary for the relationship between input and output variables to be monotonic, including both strictly monotonic and implicitly monotonic relationships. Traditional methods for maintaining monotonicity mainly rely on construction or regularization techniques, whereas this paper shows that the issue of strict monotonic probability can be viewed as a partial order between an observable revenue variable and a latent cost variable. This perspective enables us to reformulate the monotonicity challenge into modeling the latent cost variable. To tackle this, we introduce a generative network for the latent cost variable, termed the Generative Cost Model (GCM), which inherently addresses the strict monotonic problem, and propose the Implicit Generative Cost Model (IGCM) to address the implicit monotonic problem. We further validate our approach with a numerical simulation of quantile regression and conduct multiple experiments on public datasets, showing that our method significantly outperforms existing monotonic modeling techniques. The code for our experiments can be found at https://github.com/tyxaaron/GCM.

cs.LG↗

arXiv2022

Intelligent UNIT LEVEL TEST Generator for Enhanced Software Quality

Unit level test has been widely recognized as an important approach to improve the software quality, as it can expose bugs earlier during the development phase. However, manual unit level test development is often tedious and insufficient. Also, it is hard for developers to precisely identify the most error prone code block deserving the best test coverage by themselves. In this paper, we present the automatic Unit level test framework we used for intel media driver development. It can help us identify the most critical code block, provide the test coverage recommendation, and automatically generate >80% ULT code (~400K Lines of test code) as well as ~35% test cases (~7K test cases) for intel media driver. It helps us to greatly shrink the average ULT development effort from ~24 Man hours to ~3 Man hours per 1000 Lines of driver source code.

cs.SE↗

arXiv2022

Enhanced Grey Box Fuzzing For Intel Media Driver

Grey box fuzzing is one of the most successful methods for automatic vulnerability detection. However,conventional Grey box Fuzzers like AFL can open perform fuzzing against the whole input and spend more time on smaller seeds with lower execution time, which significantly impact fuzzing efficiency for complicated input types. In this work, we introduce one intelligent grey box fuzzing for Intel Media driver, MediaFuzzer, which can perform effective fuzzing based on selective fields of complicated input. Also, with one novel calling depth-based power schedule biased toward seed corpus which can lead to deeper calling chain, it dramatically improves the vulnerability exposures (~6.6 times more issues exposed) and fuzzing efficiency (~2.7 times more efficient) against the baseline AFL for Intel media driver with almost negligible overhead.

cs.CR↗

arXiv2022

Towards Maintainable Platform Software -- Delivery Cost Control in Continuous Software Development

Modern platform software delivery cost increases rapidly as it usually needs to align with many hardware and silicon's TTMs, feature evolvement and involves hundreds of engineers. In this paper, citing one ultra-large-scale software - Intel Media Driver as an example, we analyze the hotspots leading to delivery cost increase in continuous software development, the challenges on our software design and our experiences on software delivery cost shrink against the targeted design enhancements. We expect the identified hotspots can help more researchers to form the corresponding research agendas and the experiences shared can help following practitioners to apply similar enhancements.

cs.SE↗

arXiv2022

IVeri: Privacy-Preserving Interdomain Verification

In an interdomain network, autonomous systems (ASes) often establish peering agreements, so that one AS (agreement consumer) can influence the routing policies of the other AS (agreement provider). Peering agreements are implemented in the BGP configuration of the agreement provider. It is crucial to verify their implementation because one error can lead to disastrous consequences. However, the fundamental challenge for peering agreement verification is how to preserve the privacy of both ASes involved in the agreement. To this end, this paper presents IVeri, the first privacy-preserving interdomain agreement verification system. IVeri models the interdomain agreement verification problem as a SAT formula, and develops a novel, efficient, privacy-serving SAT solver, which uses oblivious shuffling and garbled circuits as the key building blocks to let the agreement consumer and provider collaboratively verify the implementation of interdomain peering agreements without exposing their private information. A prototype of IVeri is implemented and evaluated extensively. Results show that IVeri achieves accurate, privacy-preserving interdomain agreement verification with reasonable overhead.

cs.CR↗

arXiv2020

Predicting Individual Treatment Effects of Large-scale Team Competitions in a Ride-sharing Economy

Millions of drivers worldwide have enjoyed financial benefits and work schedule flexibility through a ride-sharing economy, but meanwhile they have suffered from the lack of a sense of identity and career achievement. Equipped with social identity and contest theories, financially incentivized team competitions have been an effective instrument to increase drivers' productivity, job satisfaction, and retention, and to improve revenue over cost for ride-sharing platforms. While these competitions are overall effective, the decisive factors behind the treatment effects and how they affect the outcomes of individual drivers have been largely mysterious. In this study, we analyze data collected from more than 500 large-scale team competitions organized by a leading ride-sharing platform, building machine learning models to predict individual treatment effects. Through a careful investigation of features and predictors, we are able to reduce out-sample prediction error by more than 24%. Through interpreting the best-performing models, we discover many novel and actionable insights regarding how to optimize the design and the execution of team competitions on ride-sharing platforms. A simulated analysis demonstrates that by simply changing a few contest design options, the average treatment effect of a real competition is expected to increase by as much as 26%. Our procedure and findings shed light on how to analyze and optimize large-scale online field experiments in general.

cs.CY↗

arXiv2016

k.p theory of freestanding narrow band gap semiconductor nanowires

We report on a theoretical study of the electronic structures of freestanding nanowires made from narrow band gap semiconductors GaSb, InSb and InAs. The nanowires are described by the eight-band k.p Hamiltonians and the band structures are computed by means of the finite element method in a mixture basis consisting of linear triangular elements inside the nanowires and constrained Hermite triangular elements near the boundaries. The nanowires with two crystallographic orientations, namely the [001] and [111] orientations, and with different cross-sectional shapes are considered. For each orientation, the nanowires of the three narrow band gap semiconductors are found to show qualitatively similar characteristics in the band structures. However, the nanowires oriented along the two different crystallographic directions are found to show different characteristics in the valence bands. In particular, it is found that all the conduction bands show simple, good parabolic dispersions in both the [001]- and [111]-oriented nanowires, while the top valence bands show double-maximum structures in the [001]-oriented nanowires, but single-maximum structures in the [111]-oriented nanowires. The wave functions and spinor distributions of the band states in these nanowires are also calculated. It is found that significant mixtures of electron and hole states appear in the bands of these narrow band gap semiconductor nanowires. The wave functions exhibit very different distribution patterns in the nanowires oriented along the [001] direction and the nanowires oriented along the [111] direction. It is also shown that single-band effective mass theory could not reproduce all the band state wave functions presented in this work.

cond-mat.mes-hall↗

arXiv2016

Electronic structures of free-standing nanowires made from indirect bandgap semiconductor gallium phosphide

We present a theoretical study of the electronic structures of freestanding nanowires made from gallium phosphide (GaP)--a III-V semiconductor with an indirect bulk bandgap. We consider [001]-oriented GaP nanowires with square and rectangular cross sections, and [111]-oriented GaP nanowires with hexagonal cross sections. Based on tight binding models, both the band structures and wave functions of the nanowires are calculated. For the [001]-oriented GaP nanowires, the bands show anti-crossing structures, while the bands of the [111]-oriented nanowires display crossing structures. Two minima are observed in the conduction bands, while the maximum of the valence bands is always at the $Γ$-point. Using double group theory, we analyze the symmetry properties of the lowest conduction band states and highest valence band states of GaP nanowires with different sizes and directions. The band state wave functions of the lowest conduction bands and the highest valence bands of the nanowires are evaluated by spatial probability distributions. For practical use, we fit the confinement energies of the electrons and holes in the nanowires to obtain an empirical formula.

cond-mat.mes-hall↗

arXiv2016

Topological energy gaps in the [111]-oriented InAs/GaSb and GaSb/InAs core-shell nanowires

The [111]-oriented InAs/GaSb and GaSb/InAs core-shell nanowires have been studied by the $8\times 8$ Luttinger-Kohn $\vec{k}\cdot\vec{p}$ Hamiltonian to search for non-vanishing fundamental gaps between inverted electron and hole bands. We focus on the variations of the topologically nontrivial fundamental gap, the hybridization gap, and the effective gap with the core radius and shell thickness of the nanowires. The evolutions of all the energy gaps with the structural parameters are shown to be dominantly governed by quantum size effects. With a fixed core radius, a topologically nontrivial fundamental gap exists only at intermediate shell thicknesses. The maximum gap is $\sim 4.4$ meV for GaSb/InAs and $\sim 3.5$ meV for InAs/GaSb core-shell nanowires, and for the GaSb/InAs core-shell nanowires the gap persists over a wider range of geometrical parameters. The intrinsic reason for these differences between the two types of nanowires is that in the shell the electron-like states of InAs is more delocalized than the hole-like state of GaSb, while in the core the hole-like state of GaSb is more delocalized than the electron-like state of InAs, and both features favor stronger electron-hole hybridization. Since similar features of the electron- and hole-like states have been found in nanowires of other materials, it could serve as a common rule to put the hole-like state in the core while the electron-like state in the shell of a core-shell nanowire to achieve better topological properties.

cond-mat.mes-hall↗

arXiv2015

Electronic structures of [111]-oriented free-standing InAs and InP nanowires

We report on a theoretical study of the electronic structures of the [111]-oriented, free-standing, zincblende InAs and InP nanowires with hexagonal cross sections by means of an atomistic $sp^{3}s^{*} $, spin-orbit interaction included, nearest-neighbor, tight-binding method. The band structures and the band state wave functions of these nanowires are calculated and the symmetry properties of the bands and band states are analyzed based on the $C_{3v}$ double point group. It is shown that all bands of these nanowires are doubly degenerate at the $Γ$-point and some of these bands will split into non-degenerate bands when the wave vector $k$ moves away from the $Γ$-point as a manifestation of spin-splitting due to spin-orbit interaction. It is also shown that the lower conduction bands of these nanowires all show simple parabolic dispersion relations, while the top valence bands show complex dispersion relations and band crossings. The band state wave functions are presented by the spatial probability distributions and it is found that all the band states show $2π/3$-rotation symmetric probability distributions. The effects of quantum confinement on the band structures of the [111]-oriented InAs and InP nanowires are also examined and an empirical formula for the description of quantization energies of the lowest conduction band and the highest valence band is presented. The formula can simply be used to estimate the enhancement of the band gaps of the nanowires at different sizes as a result of quantum confinement.

cond-mat.mes-hall↗