Bitcoin, as the most valuable cryptocurrency, has become a significant target for criminal activities, leading to substantial financial losses. To timely detect criminal activities and reduce losses, the accurate detection of illicit addresses is crucial. Current research typically involves extracting features from addresses to classify them as either licit or illicit. However, they neglect the transaction topology information associated with the addresses, thereby compromising the efficiency. In this article, we focus on improving the utilization of address information and propose DIBA detector, an effective framework designed for the automatic D etection of I llicit B itcoin A ddresses. DIBA first incorporates a transaction graph construction module that constructs per-address transaction graph based on UTXO model, thereby mapping to the transaction topology for each address. Subsequently, a novel hybrid spatiotemporal network is designed to learn graph representation for each per-address transaction graph, generating comprehensive graph embeddings that serve as critical inputs for the final classification model. Experimental results demonstrate that our proposed framework outperforms existing state-of-the-art methods for detecting illicit Bitcoin addresses, achieving precision and F1-score values of 92.77% and 93.58%, respectively. As a side contribution, we construct a dataset comprising over 180,000 addresses, with half labeled as licit and half as illicit. Our dataset is released at https://github.com/dlvlb123/DIBA .

Deep neural network-based website fingerprinting (WF) attacks have significantly threatened anonymous communication systems, like The Onion Router (Tor). WF defense methods based on adversarial learning techniques offer potential resistance against deep WF attacks. However, existing defenses primarily focus on packet-level perturbations, often neglecting fine-grained perturbations at the packet interval-level. In this paper, we present a novel WF defense method, called Amoeba, which is designed to generate traces that incorporate both packet-level and packet interval-level perturbations to defend against deep WF attacks while maintaining controllable overhead. Amoeba employs a two-stage framework: in the first stage, it learns the trace patterns by adversarial learning techniques with Wasserstein Generative Adversarial Networks (WGAN). The second stage introduces a dynamic cost control mechanism to pad dummy packets into normal traces to perturb trace patterns with low bandwidth overhead and time costs. Furthermore, we collect a new Tor trace dataset with Tor anonymous communication system and conduct extensive experiments on both a public dataset and our collected dataset. Experimental results demonstrate that Amoeba can reduce the classification accuracy of deep WF attacks from over 90% to below 40%, and outperform existing defenses. Moreover, Amoeba has over 4% bandwidth overhead and 13% time costs reduction than existing defenses, which shows the effectiveness and superiority of Amoeba in resisting deep WF attacks compared to existing methods.

Abstract Website fingerprinting (WF) attacks based on deep neural networks can effectively identify the target website. Recently, WF defence methods with trace adversarial examples (DTAE) can reduce classification accuracy of existing deep WF attacks from over 98% to approximately 50%. To overcome the vulnerability of deep WF attacks in classifying traces defended by adversarial examples, we propose a novel WF attack method specially designed for DTAE, called multiscale fingerprinting (MF). Specifically, MF slices each trace with different time slots to build a multi-channel matrix as the model input, and utilizes a multiscale convolutional neural network to extract the real trace patterns, which are changed and complicated by DTAE. Furthermore, we evaluate the performance of MF in closed- and open-world scenarios on the public dataset. In the closed-world scenario, MF achieves superior performance on DTAE methods, exhibiting an improvement of 12.32% over the state-of-the-art DTAE. In the open-world scenario, MF attains better performance on defended traces on Recall and Precision metrics, which demonstrates the enhanced robustness compared to existing deep WF attacks.

In this paper, we address the 3D position control of RCM-constrained instruments with monocular cameras. To compute the instrument's position from a single 2D image, we develop an innovative gradient descent algorithm which rotates and translates a line segment (over the plane spanned by the imaged instrument and the optical centre) until it best aligns with the manipulated tool. In contrast with other approaches in the literature, our algorithm only requires to simultaneously observe two feature points; the proposed iterative algorithm is not based on the exact solution, therefore it can still work with noisy image measurements. We derive a kinematic controller that uses the proposed position estimator to guide the 3D motion of a robotic instrument with a monocular camera. We evaluate the performance of our approach with numerical simulations and experiments.