Nov 2026· Proceedings of the 2026 ACM SIGSAC Conference on Computer and Communications Security· 0 citations· 14 references
TL;DR
NeuGo, a directed fuzzing framework for Adversarial Neuron Fuzzing that targets operator-level execution semantics that identifies numerically unstable operator states via constraint-aware sampling and sensitivity analysis, applies gradient-based optimization to synthesize model inputs that guide execution into these latent defects.
Abstract
Deep learning systems depend on a complex tool chain of software libraries and hardware backends. This complexity improves performance but also creates an under-explored attack surface: implementation precision bugs whose numerical instability can be exploited adversarially rather than treated as benign noise. We formalize this threat as Code Adversarial Attacks (CAAs), in which operator implementations exhibit divergent numerical behavior across execution environments, enabling supply-chain Trojan operators that behave correctly under standard environment but induce targeted failures in specific deployment environment. To detect this type of threats, we introduce NeuGo, a directed fuzzing framework for Adversarial Neuron Fuzzing that targets operator-level execution semantics. NeuGo focuses on internal operator-level states rather than model outputs: it (i) identifies numerically unstable operator states via constraint-aware sampling and sensitivity analysis, and (ii) applies gradient-based optimization to synthesize model inputs that guide execution into these latent defects, further confirmed with differential testing across multiple software-hardware configurations. We evaluated NeuGo on over 100 production models across sixteen execution environments, including PyTorch, TensorFlow, and ONNX Runtime on CPU, GPU, and XPU platforms. NeuGo identifies more than 30 reproducible implementation bugs in these widely-used frameworks and runtimes. We further demonstrate the security implications of these bugs by realizing supply-chain backdoors in the form of Trojan operators that evade conventional validation yet cause misclassifications on targeted deployment hardware. Our results show that implementation precision bugs constitute a concrete security threat for deployed deep learning systems, necessitating rigorous integrity checks.
This publication proposes a definition and a classification of agile software development approaches and analyses ten software development methods that can be characterized as being "agile" against the defined criterion.
P. Abrahamsson, O. Salo, Jussi Ronkainen et al.· arXiv.org· 727 citations· ⚡54
The study shows that agile practices improve both informal and formal communication, but indicates that, in larger development situations involving multiple external stakeholders, a mismatch of adequate communication mechanisms can sometimes even hinder the communication.
M. Pikkarainen, Jukka Haikara, O. Salo et al.· Empirical Software Engineeri...· 401 citations· ⚡48
The results indicate that software engineering work practices are chosen opportunistically, adapted and configured to provide value under the constrains imposed by the startup context.
Nicolò Paternoster, Carmine Giardino, M. Unterkalmsteiner et al.· Information and Software Tec...· 394 citations· ⚡54
The results show that the embedded industry has been able to apply agile methods in its development processes and that the appreciation of the agile methods and their individual practices appears to increase once adopted and applied in practice.
O. Salo, P. Abrahamsson· IET Software· 238 citations· ⚡9
Consequences of happiness and unhappiness that are beneficial and detrimental for developers' mental well-being, the software development process, and the produced artifacts are found.
D. Graziotin, Fabian Fagerholm, Xiaofeng Wang et al.· Journal of Systems and Softw...· 236 citations· ⚡13
The Mobile-D approach is briefly outlined here and the experiences gained from four case studies are discussed, which helped develop an agile development approach for mobile application development.
P. Abrahamsson, Antti Hanhineva, H. Hulkko et al.· Conference on Object-Oriente...· 225 citations· ⚡18
Related blog posts
Microsoft Research Blog· microsoft.comNov 12, 2026
What's new in research at Microsoft? Host Anson Ho explores the latest, including AI-powered universal translation, a breakthrough in a 31-year-old mathematics challenge, and advances making AI coding agents faster and more efficient. The post Microsoft Research Focus – Edition #98 appeared first on Microsoft Research.
AI agents can generate code at a scale that’s hard for humans to review line-by-line. PDD Kit helps ensure the code is doing what the developer intended by creating formal specifications and testing against them. It converts developer intent into a Verus specification, proves AI-generated code meets it for every input, and uses tests where proofs aren't practical. Developers review intent, not every line. The post Formally verifying agent-written full-stack web apps with PDD Kit appeared first o…