Search for: All records

Imaging nanomaterials in hybrid systems is critical to understanding the structure and functionality of these systems. However, current technologies such as scanning electron microscopy (SEM) may obtain high resolution/contrast images at the cost of damaging or contaminating the sample. For example, to prevent the charging of organic substrate/matrix, a very thin layer of metal is coated on the surface, which will permanently contaminate the sample and eliminate the possibility of reusing it for following processes. Conversely, examining the sample without any modifications, in pursuit of high-fidelity digital images of its unperturbed state, can come at the cost of low-quality images that are challenging to process. Here, a solution is proposed for the case where no rightness threshold is available to reliably judge whether a region is covered with nanomaterials. The method examines local brightness variability to detect nanomaterial deposits. Very good agreement with manually obtained values of the coverage is observed, and a strong case is made for the method’s automatability. Although the developed methodology is showcased in the context of SEM images of Polydimethylsiloxane (PDMS) substrates on which silicone dioxide (SiO2) nanoparticles are assembled, the underlying concepts may be extended to situations where straightforward brightness thresholding is not viable. 

Graph-based anomaly detection is pivotal in diverse security applications, such as fraud detection in transaction networks and intrusion detection for network traffic. Standard approaches, including Graph Neural Networks (GNNs), often struggle to generalize across shifting data distributions. For instance, we observe that a real-world eBay transaction dataset revealed an over 50% decline in fraud detection accuracy when adding data from only a single new day to the graph due to data distribution shifts. This highlights a critical vulnerability in purely data-driven approaches. Meanwhile, real-world domain knowledge, such as "simultaneous transactions in two locations are suspicious," is more stable and a common existing component of real-world detection strategies. To explicitly integrate such knowledge into data-driven models such as GCNs, we propose KnowGraph, which integrates domain knowledge with data-driven learning for enhanced graph-based anomaly detection. KnowGraph comprises two principal components: (1) a statistical learning component that utilizes a main model for the overarching detection task, augmented by multiple specialized knowledge models that predict domain-specific semantic entities; (2) a reasoning component that employs probabilistic graphical models to execute logical inferences based on model outputs, encoding domain knowledge through weighted first-order logic formulas. In addition, KnowGraph has leveraged the Predictability-Computability-Stability (PCS) framework for veridical data science to estimate and mitigate prediction uncertainties. Empirically, KnowGraph has been rigorously evaluated on two significant real-world scenarios: collusion detection in the online marketplace eBay and intrusion detection within enterprise networks. Extensive experiments on these large-scale real-world datasets show that KnowGraph consistently outperforms state-of-the-art baselines in both transductive and inductive settings, achieving substantial gains in average precision when generalizing to completely unseen test graphs. Further ablation studies demonstrate the effectiveness of the proposed reasoning component in improving detection performance, especially under extreme class imbalance. These results highlight the potential of integrating domain knowledge into data-driven models for high-stakes, graph-based security applications. 

Abstract Bacterial colonies growing on solid surfaces can exhibit robust expansion kinetics, with constant radial growth and saturating vertical expansion, suggesting a common developmental program. Here, we study this process forEscherichia colicells using a combination of modeling and experiments. We show that linear radial colony expansion is set by the verticalization of interior cells due to mechanical constraints rather than radial nutrient gradients as commonly assumed. In contrast, vertical expansion slows down from an initial linear regime even while radial expansion continues linearly. This vertical slowdown is due to limitation of cell growth caused by vertical nutrient gradients, exacerbated by concurrent oxygen depletion. Starvation in the colony interior results in a distinct death zone which sets in as vertical expansion slows down, with the death zone increasing in size along with the expanding colony. Thus, our study reveals complex heterogeneity within simple monoclonal bacterial colonies, especially along the vertical dimension. The intricate dynamics of such emergent behavior can be understood quantitatively from an interplay of mechanical constraints and nutrient gradients arising from obligatory metabolic processes. 

With the rapidly increasing capabilities and adoption of code agents for AI-assisted coding and software development, safety and security concerns, such as generating or executing malicious code, have become significant barriers to the real-world deployment of these agents. To provide comprehensive and practical evaluations on the safety of code agents, we propose RedCode, an evaluation platform with benchmarks grounded in four key principles: real interaction with systems, holistic evaluation of unsafe code generation and execution, diverse input formats, and high-quality safety scenarios and tests. RedCode consists of two parts to evaluate agents’ safety in unsafe code execution and generation: (1) RedCode-Exec provides challenging code prompts in Python as inputs, aiming to evaluate code agents’ ability to recognize and handle unsafe code. We then map the Python code to other programming languages (e.g., Bash) and natural text summaries or descriptions for evaluation, leading to a total of over 4,000 testing instances. We provide 25 types of critical vulnerabilities spanning various domains, such as websites, file systems, and operating systems. We provide a Docker sandbox environment to evaluate the execution capabilities of code agents and design corresponding evaluation metrics to assess their execution results. (2) RedCode-Gen provides 160 prompts with function signatures and docstrings as input to assess whether code agents will follow instructions to generate harmful code or software. Our empirical findings, derived from evaluating three agent frameworks based on 19 LLMs, provide insights into code agents’ vulnerabilities. For instance, evaluations on RedCode-Exec show that agents are more likely to reject executing unsafe operations on the operating system, but are less likely to reject executing technically buggy code, indicating high risks. Unsafe operations described in natural text lead to a lower rejection rate than those in code format. Additionally, evaluations on RedCode-Gen reveal that more capable base models and agents with stronger overall coding abilities, such as GPT4, tend to produce more sophisticated and effective harmful software. Our findings highlight the need for stringent safety evaluations for diverse code agents. Our dataset and code are publicly available at https://github.com/AI-secure/RedCode. 

Abstract Seismic arrays constrain local wave propagation that can be used to infer earthquake source characteristics. Array processing is routinely used to infer detailed earthquake properties of intermediate and large events. However, the source properties of microseismicity often remain elusive. In this study, we use high signal-to-noise ratio seismograms of 204 ML 0.0–1.8 earthquakes induced by the 6 km deep 2018 Espoo/Helsinki geothermal stimulation to evaluate the performance and capabilities of beamforming and backprojection array methods. Using accurate travel-time-based event locations as a reference, we first show that miniarray beamforming is sensitive to medium heterogeneities and requires calibration to mitigate local systematic slowness biases. A catalog-based calibration significantly improves our multiarray beam raytracing estimates of source locations. Second, the application of the backprojection technique using P-wave signals with sufficient azimuthal coverage yields hypocenter estimates with generally good horizontal but poor vertical resolution. The short local source–receiver distances result in incomplete separation of P- and S-wave arrivals during backprojection. Numerical tests show that the relatively large S-wave amplitudes can influence coherent P-wave stacks, resulting in large location errors. Our combined P- and S-wave backprojection approach mitigates the influence of the large S-wave amplitude and improves the depth resolution significantly. The average depth offset to the reference catalog locations reduces from ≥1.4 km to ∼91 m. Third, 3D numerical simulations demonstrate that backprojection swimming patterns are not merely processing or configuration artifacts. We show that the swimming patterns correlate with and can resolve the source focal mechanism when the azimuthal wavefield sampling is sufficiently complete. Our work demonstrates that the backprojection techniques can help to better constrain important properties of local-scale microseismicity.