In supervised machine learning, use of correct labels is extremely important to ensure high accuracy. Unfortunately, most datasets contain corrupted labels. Machine learning models trained on such do not generalize well. Thus, detecting the label errors can significantly increase their efficacy. We propose a novel framework, called CTRL <xref ref-type="fn" rid="fn1" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">¹</xref> <fn id="fn1"...

In the past decade, security of cellular networks has been increasingly under scrutiny, leading to discovery numerous vulnerabilities that expose network and its users a wide range risks, from denial service information leak. However, most these findings have made through ad-hoc manual analysis, which is inadequate for fundamentally enhancing assurance system as complex network. An important observation massive amount technical documentation can provide key insights into protection it puts...

We present a novel reduced-order fluid simulation technique leveraging Dynamic Mode Decomposition (DMD) to achieve fast, memory-efficient, and user-controllable subspace simulation. demonstrate that our approach combines the strengths of both spatial reduced order models (ROMs) as well spectral decompositions. By optimizing for operator evolves system state from one timestep next, rather than itself, we gain compressive power ROMs intuitive physical dynamics methods. The latter property is...

AODV on-demand routing protocol is widely deployed in ad hoc network, but it has some drawbacks. This paper proposes an improved called AODV-I. By adding the congestion processing to RREQ message and repair mechanism message, new not only reduces packet loss rate end-to-end latency, also enhances utilization of network resources. Finally, we analyze performance through simulation experiment.

The long runtime of high-fidelity partial differential equation (PDE) solvers makes them unsuitable for time-critical applications. We propose to accelerate PDE using reduced-order modeling (ROM). Whereas prior ROM approaches reduce the dimensionality discretized vector fields, our continuous (CROM) approach builds a low-dimensional embedding fields themselves, not their discretization. represent this reduced manifold continuously differentiable neural which may train on any and all...

As the large amount of video surveillance data floods into cloud center, achieving load balancing in a network has become challenging problem. Meanwhile, we hope center maintains low latency, consumption, and high throughput performance when transmitting massive amounts data. OpenFlow enables software-defined solution through programing to control scheduling flow center. However, existing algorithm cannot cope with congestion effectively. Even for some dynamic algorithms, adjustments can...

In supervised machine learning, use of correct labels is extremely important to ensure high accuracy. Unfortunately, most datasets contain corrupted labels. Machine learning models trained on such do not generalize well. Thus, detecting their label errors can significantly increase efficacy. We propose a novel framework, called CTRL (Clustering TRaining Losses for error detection), detect in multi-class datasets. It detects two steps based the observation that learn clean and noisy different...

Recent years have witnessed tremendous success in Self-Supervised Learning (SSL), which has been widely utilized to facilitate various downstream tasks Computer Vision (CV) and Natural Language Processing (NLP) domains.However, attackers may steal such SSL models commercialize them for profit, making it crucial verify the ownership of models.Most existing protection solutions (e.g., backdoorbased watermarks) are designed supervised learning cannot be used directly since they require that...

The Internet of Things (IoT) is rapidly transforming our lives and work, enabling a wide range emerging services applications. However, as the scale IoT expands, its security issues are becoming increasingly prominent. Malicious actors can exploit vulnerabilities in devices to launch attacks. Protecting begins with device identification. Identified have corresponding protective measures selected based on information, thereby enhancing network security. In this study, we propose...

We present a neural operator architecture to simulate Lagrangian dynamics, such as fluid flow, granular flows, and elastoplasticity. Traditional numerical methods, the finite element method (FEM), suffer from long run times large memory consumption. On other hand, approaches based on graph networks are faster but still computation dense graphs, which often required for high-fidelity simulations. Our model, GIOROM or Graph Interaction Operator Reduced-Order Modeling, learns temporal dynamics...

The ubiquity of neural networks (NNs) in real-world applications, from healthcare to natural language processing, underscores their immense utility capturing complex relationships within high-dimensional data. However, NNs come with notable disadvantages, such as "black-box" nature, which hampers interpretability, well tendency overfit the training We introduce a novel method for learning interpretable differentiable logic (DLNs) that are architectures employ multiple layers binary...

Recent years have witnessed tremendous success in Self-Supervised Learning (SSL), which has been widely utilized to facilitate various downstream tasks Computer Vision (CV) and Natural Language Processing (NLP) domains. However, attackers may steal such SSL models commercialize them for profit, making it crucial verify the ownership of models. Most existing protection solutions (e.g., backdoor-based watermarks) are designed supervised learning cannot be used directly since they require that...

New lumping kinetic models, considering the effects of nitrogen content in product and correction coefficient LHSV, were proposed to describe hydrodesulfurization crude Longkou shale oil. The parameters obtained using nonlinear regression experimental data which conducted a bench-scale trick-bed reactor with NiW/Al 2 O 3 catalyst at various conditions. results show that 4-lump model is optimal model. values apparent activation energies lumps 1, 2, 4 are 51.14, 62.64, 130 166.42kJ/mol,...

The hydrodenitrogenation (HDN) of Longkou shale oil were carried out in microscale trick-bed reactor over a commercial NiW/Al 2 O 3 catalyst with high HDN activity. effects temperature, pressure, liquid hourly space velocity (LHSV) and hydrogen/oil ratio on the conversions total, basic non-basic nitrogen compounds investigated at various conditions (340-420°C, 0.2-2h -1 , 4-9MPa 400-1000 L/L). results show that have higher reactivity than nitrogen. distributions species also investigated....

With the rapid popularization of Internet and development network technology, people are increasingly inclined to online shopping. This paper uses B/S three-layer structure, JSP technology for dynamic page design. From perspective system security code reusability, JavaBean is used encapsulate key program. The background database MySQL database. "SanWei" e-commerce bookstore platform includes foreground management management, mainly including shopping cart commodity search, user data...

Linear reduced-order modeling (ROM) simplifies complex simulations by approximating the behavior of a system using simplified kinematic representation. Typically, ROM is trained on input created with specific spatial discretization, and then serves to accelerate same discretization. This discretization-dependence restrictive. Becoming independent discretization would provide flexibility mix match mesh resolutions, connectivity, type (tetrahedral, hexahedral) in training data; novel...