Unmasking Bitcoin: A New Approach to Transaction Tracing
Researchers have developed novel methods for identifying the origins of Bitcoin transactions by analyzing network traffic patterns.
Science
Outsmarting Attackers: Adaptive Cyber Deception for Critical Infrastructure
A new framework leverages artificial intelligence to dynamically defend industrial control systems against evolving cyber threats.
Guarding the Source: Detecting Malicious Content in AI-Powered Search
New research explores how to identify harmful or misleading information injected into the knowledge base of systems that augment language models with external data.
Decoding Option Prices with Simplicity
New research reveals that surprisingly simple neural networks can accurately capture the complex information embedded in option prices and implied volatility surfaces.
Decoding the Unexpected: A New Approach to Anomaly Detection
Researchers have developed a novel method for identifying unusual data points by examining the internal workings of existing neural networks.
Beyond ‘Why’: Crafting AI Explanations People Actually Understand
New research reveals that simply explaining how an AI made a decision isn’t enough – effective explanations need to provide supporting details and hidden reasoning.
The AI Image Authenticity Arms Race: A New Framework for Detection
Researchers have unveiled EvoGuard, an adaptable system that uses intelligent agents and diverse detection methods to stay ahead of increasingly sophisticated AI-generated imagery.
Can You Spot the Bot?
![The study constructs paired corpora of human and large language model outputs - [latex]HC3[/latex] (23k pairs) and [latex]ELI5[/latex] (15k pairs) - to benchmark three detector families-classical statistical classifiers, fine-tuned encoder transformers (including BERT, RoBERTa, and DeBERTa-v3), and large language models prompted as detectors-assessing their ability to distinguish between human and machine-generated text through a unified five-metric suite, and further explores generalization across models alongside adversarial techniques to humanize machine outputs at multiple levels.](https://arxiv.org/html/2603.17522v1/figures/model_architecture.png)
A new study rigorously tests the accuracy of tools designed to identify text written by artificial intelligence.
Beyond Black Boxes: Smarter Transformers for Financial Forecasting
A new distillation framework injects expert knowledge into Transformer models, boosting accuracy and resilience in volatile financial markets.
Fortifying Network Defenses with AI and Synthetic Data
![Generative Adversarial Networks (GANs) and Wasserstein GANs (WGANs) demonstrate distinct loss progress characteristics during training, with WGANs exhibiting more stable convergence due to their use of the Earth Mover’s distance [latex] E(X,Y) = \in f_{γ ∈ Π(X,Y)} E_{x∼X, y∼Y}[c(x,y)] [/latex] as a loss function.](https://arxiv.org/html/2603.17717v1/fig2.png)
This review explores how machine learning can enhance intrusion detection systems and investigates the potential of artificially generated data to improve their accuracy and resilience.