Transfer learning, a method that influences pretrained models to improve performance on new tasks, faces several data privacy, security, and scalability challenges when applied in decentralized environments. Recently, there have been few attempts to integrate blockchain technology with a transfer-learning approach. However, these classical blockchainbased non-collaborative transfer learning models face significant challenges, including poor scalability, security vulnerability, lack of data diversity, and inefficient incentive structures, thereby hindering the effectiveness and efficiency of the transfer learning process. This paper presents a unified framework that utilizes decentralized blockchain technology for collaborative transfer learning. The framework enables secure model sharing within a collaborative learning environment, supported by incentive mechanisms while maintaining model integrity and providing context protection. We perform extensive experimental studies with eight well-known large-scale and fine-grained image datasets. After initial transfer learning, incorporating collaborative learning via the decentralized blockchain network yielded promising results. The global model, after aggregating contributions from all entities, shows an impressive 6.8 % to 10.0 % accuracy improvement compared to individual baseline models. For blockchain implementation, we use the Rahasak blockchain as the ledger and its Aplos platform for a customized smart contract interface. Rahasak-CA, the certificate authority of the Rahasak blockchain, stores the digital certificates of peers in the transfer learning process. The system was deployed using Docker and Kubernetes, and evaluated on the proposed testbed.

Machine Learning as a Service (MLaaS) enables users to leverage powerful machine learning models through cloud-based APIs, offering scalability and ease of deployment. However, these services are vulnerable to model extraction attacks, where adversaries repeatedly query the application programming interface (API) to reconstruct a functionally similar model, compromising intellectual property and security. Despite various defense strategies being proposed, many suffer from high computational costs, limited adaptability to evolving attack techniques, and a reduction in performance for legitimate users. In this paper, we introduce a Resilient Adaptive Defense Framework for Model Extraction Attack Protection (RADEP), a multifaceted defense framework designed to counteract model extraction attacks through a multi-layered security approach. RADEP employs progressive adversarial training to enhance model resilience against extraction attempts. Malicious query detection is achieved through a combination of uncertainty quantification and behavioral pattern analysis, effectively identifying adversarial queries. Furthermore, we develop an adaptive response mechanism that dynamically modifies query outputs based on their suspicion scores, reducing the utility of stolen models. Finally, ownership verification is enforced through embedded watermarking and backdoor triggers, enabling reliable identification of unauthorized model use. Experimental evaluations demonstrate that RADEP significantly reduces extraction success rates while maintaining high detection accuracy with minimal impact on legitimate queries. Extensive experiments show that RADEP effectively defends against model extraction attacks and remains resilient even against adaptive adversaries, making it a reliable security framework for MLaaS models.

Emergency vehicles (EMVs) play a crucial role in responding to time-critical events such as medical emergencies and fire outbreaks in an urban area. The less time EMVs spend traveling through the traffic, the more likely it would help save people's lives and reduce property loss. To reduce the travel time of EMVs, prior work has used route optimization based on historical traffic-flow data and traffic signal pre-emption based on the optimal route. However, traffic signal pre-emption dynamically changes the traffic flow which, in turn, modifies the optimal route of an EMV. In addition, traffic signal pre-emption practices usually lead to significant disturbances in traffic flow and subsequently increase the travel time for non-EMVs. In this paper, we propose EMVLight, a decentralized reinforcement learning (RL) framework for simultaneous dynamic routing and traffic signal control. EMVLight extends Dijkstra's algorithm to efficiently update the optimal route for the EMVs in real-time as it travels through the traffic network. The decentralized RL agents learn network-level cooperative traffic signal phase strategies that not only reduce EMV travel time but also reduce the average travel time of non-EMVs in the network. This benefit has been demonstrated through comprehensive experiments with synthetic and real-world maps. These experiments show that EMVLight outperforms benchmark transportation engineering techniques and existing RL-based signal control methods.

The incorporation of appropriate inductive bias plays a critical role in learning dynamics from data. A growing body of work has been exploring ways to enforce energy conservation in the learned dynamics by encoding Lagrangian or Hamiltonian dynamics into the neural network architecture. These existing approaches are based on differential equations, which do not allow discontinuity in the states and thereby limit the class of systems one can learn. However, in reality, most physical systems, such as legged robots and robotic manipulators, involve contacts and collisions, which introduce discontinuities in the states. In this paper, we introduce a differentiable contact model, which can capture contact mechanics: frictionless/frictional, as well as elastic/inelastic. This model can also accommodate inequality constraints, such as limits on the joint angles. The proposed contact model extends the scope of Lagrangian and Hamiltonian neural networks by allowing simultaneous learning of contact and system properties. We demonstrate this framework on a series of challenging 2D and 3D physical systems with different coefficients of restitution and friction. The learned dynamics can be used as a differentiable physics simulator for downstream gradient-based optimization tasks, such as planning and control.

We propose a novel network architecture, namely InverseNet, to solve the inverse problems of multimedia data. The inverse problem is cast in the form of learning an end-to-end mapping from observed multimedia data to the ground-truth data. Inspired by the splitting strategy to tackle inverse problems, the mapping is learned by InverseNet, a composition of two networks, with one handling the inversion of the physical forward model and the other handling the denoising of the output from the former network. Training InverseNet is annealing as the intermediate variable between these two networks bridges the gap between the input and output and progressively approaches to the ground-truth. Extensive experiments on synthetic and real multimedia datasets on the tasks, e.g., motion deblurring, super-resolution, and colorization, demonstrate the efficiency and accuracy of the proposed method compared with other image processing algorithms.

Deep generative models can perform dramatically better than traditional graphical models in a number of machine learning tasks. However, training such models remains challenging because their latent variables typically do not have an analytical posterior distribution, largely due to the nonlinear activation nodes. We present a deep convolutional factor analyser (DCFA) for multivariate time series modeling. Our network is constructed in a way that bottom layer nodes are independent. Through a process of up-sampling and convolution, higher layer nodes gain more temporal dependency. Our model can thus give a time series different representations at different depths. DCFA only consists of linear Gaussian nodes. Therefore, the posterior distributions of latent variables are also Gaussian and can be estimated easily using standard variational Bayes algorithm. We show that even without nonlinearity the proposed deep model can achieve state-of-the-art results in anomaly detection, classification and clustering using both synthetic and real-world datasets.

This work considers the optimal scheduling problem for a campus central plant equipped with a bank of multiple electrical chillers and a thermal energy storage (TES). Typically, the chillers are operated in ON/OFF modes to charge TES and supply chilled water to satisfy the campus cooling demands. A bilinear model is established to describe the system dynamics of the central plant. A model predictive control (MPC) problem is formulated to obtain optimal set-points to satisfy the campus cooling demands and minimize daily electricity cost. At each time step, the MPC problem is represented as a large-scale mixed-integer nonlinear programming problem. We propose a heuristic algorithm to obtain suboptimal solutions for it via dynamic programming (DP) and mixed integer linear programming (MILP). The system dynamics is linearized along the simulated trajectories of the system. The optimal TES operation profile is obtained by solving a DP problem at every horizon, and the optimal chiller operations are obtained by solving an MILP problem at every time step with a fixed TES operation profile. Simulation results show desired performance and computational tractability of the proposed algorithm. This work was motivated by the supervisory control need for a campus central plant. Plant operators have to decide a scheduling strategy to mix and match various chillers with a thermal energy storage to satisfy the campus cooling demands, while minimizing the operation cost. This work mathematically characterizes the system dynamics of a campus central plant and establishes a linear model to predict campus cooling load. It proposes a model predictive control (MPC) strategy to optimally schedule the campus central plant based on plant system dynamics and predicted campus cooling load. A heuristic algorithm is proposed to obtain suboptimal solutions for the MPC problem. The effectiveness and efficiency of the proposed approach are well demonstrated for the central plant at the University of California, Irvine.

Abstract This paper proposes a novel application of topic models to do entity relation detection (ERD). In order to make use of the latent semantics of text, we formulate the task of relation detection as a topic modeling problem. The motivation is to find underlying topics that are indicative of relations between named entities (NEs). Our approach considers pairs of NEs and features associated with them as mini documents, and aims to utilize the underlying topic distributions as indicators for the types of relations that may exist between the NE pair. Our system, ERD-MedLDA, adapts Maximum Entropy Discriminant Latent Dirichlet Allocation (MedLDA) with mixed membership for relation detection. By using supervision, ERD-MedLDA is able to learn topic distributions indicative of relation types. Further, ERD-MedLDA is a topic model that combines the benefits of both, maximum likelihood estimation (MLE) and maximum margin estimation (MME), and the mixed-membership formulation enables the system to incorporate heterogeneous features. We incorporate different features into the system and perform experiments on the ACE 2005 corpus. Our approach achieves better overall performance for precision, recall, and F-measure metrics as compared to baseline SVM-based and LDA-based models. We also find that our system shows better and consistent improvements with the addition of complex informative features as compared to baseline systems.

Compressed sensing, an emerging multidisciplinary field involving mathematics, probability, optimization, and signal processing, focuses on reconstructing an unknown signal from a very limited number of samples. Because information such as boundaries of organs is very sparse in most MR images, compressed sensing makes it possible to reconstruct the same MR image from a very limited set of measurements significantly reducing the MRI scan duration. In order to do that however, one has to solve the difficult problem of minimizing nonsmooth functions on large data sets. To handle this, we propose an efficient algorithm that jointly minimizes the ℓ1norm, total variation, and a least squares measure, one of the most powerful models for compressive MR imaging. Our algorithm is based upon an iterative operator-splitting framework. The calculations are accelerated by continuation and takes advantage of fast wavelet and Fourier transforms enabling our code to process MR images from actual real life applications. We show that faithful MR images can be reconstructed from a subset that represents a mere 20 percent of the complete set of measurements.

MOTIVATION: Background correction is an important preprocess in cDNA microarray data analysis. A variety of methods have been used for this purpose. However, many kinds of backgrounds, especially inhomogeneous ones, cannot be estimated correctly using any of the existing methods. In this paper, we propose the use of the TV+L1 model, which minimizes the total variation (TV) of the image subject to an L1-fidelity term, to correct background bias. We demonstrate its advantages over the existing methods by both analytically discussing its properties and numerically comparing it with morphological opening. RESULTS: Experimental results on both synthetic data and real microarray images demonstrate that the TV+L1 model gives the restored intensity that is closer to the true data than morphological opening. As a result, this method can serve an important role in the preprocessing of cDNA microarray data.

Document processing and understanding is important for a variety of applications such as office automation, creation of electronic manuals, online documentation and annotation etc. The first step towards this process often involves the extraction of relevant keywords and phrases from the documents so that they can be automatically hyperlinked within and outside the document so that we can create an electronic document. This paper describes a novel method for extracting anchorable information units (AIUs), also known as hotspots from any type of portable document format (PDF) files, which may either be created using either an editor or by scanning in documents. The AIUs are used to make these documents more intelligent for content cross-referencing to/from related multimedia documents within an electronic document publishing environment. Domain specific knowledge about the documents are used to aid the extraction process. Once the location and extent of the texts are found, the content is extracted through the use of an optical character recognition (OCR) software if necessary. For the case of object extraction for highlighting, first the images are extracted and then a variety of image processing algorithms are applied.

The product training manuals for complex machinery often consist of large-scale presentation-based documents with rich media. The content of the manuals is typically required to use up-to-date product information for effective training. Current training authoring tools have limitations in creating such large-scale product training course content in a timely manner due to the labor-intensive and time-consuming authoring processes. This paper presents an automated method to improve the process. It is based on formal specifications to dynamically generate DSSSL document styles for transforming the existing storage-based product documents into large-scale presentation-based product training manuals. The generated manuals are based on SCORM, an industry training course specification.

Document processing and understanding is important for a variety of applications such as office automation, creation of electronic manuals, online documentation and annotation etc. However the understanding and identifying of graphical objects in large engineering drawings which often can be of the order of 10000/spl times/10000 pixels using traditional methods can be a very challenging and time consuming task due to the sheer size. We describe a method that circumvents that by using a hierarchical top-down approach for recognizing graphical objects. The speedup is almost of the order of 1000 and this is achieved by following a strategy that exploits the characteristics of such diagrams such as low pixel density.

Robust segmentation of structures from an image is essential for a variety of image analysis problems. However, the conventional methods of region-based segmentation and gradient-based boundary finding are often frustrated by poor image quality. Here we propose a method to integrate the two approaches using game theory in an effort to form a unified approach that is robust to noise and poor initialization. This combines the perceptual notions of complete boundary information using edge data and shape priors with gray-level homogeneity using two computational modules. The novelty of the method is that this is a bidirectional framework, whereby both computational modules improve their results through mutual information sharing. A number of experiments were performed both on synthetic datasets and datasets of real images to evaluate the new approach and it is shown that the integrated method typically performs better than conventional gradient-based boundary finding.

The wide availability of high resolution magnetic resonance images (MRI) of the brain has facilitated tremendous progress in neuroscience. Accurate automated segmentation and quantification of neuroanatomical structure from such images is crucial for the advancement of the understanding of brain morphology, both in normal variation and in disease. Gradient-based deformable surface finding is a powerful technique for locating structure in three-dimensional images. However, it often suffers from poorly defined edges and noise. This paper proposes a gradient-based deformable surface finding approach that integrates region information. This makes the resulting procedure more robust to noise and improper initialization. In addition, prior shape information may be incorporated. The algorithm uses Gauss's Divergence theorem to find the surface of a homogeneous region-classified area in the image and integrates this with a gray-level gradient-based surface finder. Experimental results on synthetic and MR brain images show that a significant improvement is achieved as a consequence of the use of this extra information. Further, these improvements are achieved with little increase in computational overhead, an advantage derived from the application of Gauss's Divergence theorem.

Accurately segmenting and quantifying structures is a key issue in biomedical image analysis. The two conventional methods of image segmentation, region-based segmentation, and boundary finding, often suffer from a variety of limitations. Here the authors propose a method which endeavors to integrate the two approaches in an effort to form a unified approach that is robust to noise and poor initialization. The authors' approach uses Green's theorem to derive the boundary of a homogeneous region-classified area in the image and integrates this with a gray level gradient-based boundary finder. This combines the perceptual notions of edge/shape information with gray level homogeneity. A number of experiments were performed both on synthetic and real medical images of the brain and heart to evaluate the new approach, and it is shown that the integrated method typically performs better when compared to conventional gradient-based deformable boundary finding. Further, this method yields these improvements with little increase in computational overhead, an advantage derived from the application of the Green's theorem.

Precise segmentation of underlying objects in an image is very important especially for biomedical image analysis. We present an integrated approach for boundary finding using region and curvature information along with the gradient. Unlike the previous methods, where smoothing is enforced by penalizing curvature, here the grey level curvature is used as an extra source of information. However, information fusion may not be useful unless used properly. To address that, we present results that highlight the pros and cons of using the various sources of information and indicate when one should get precedence over the others.>

Accurately segmenting and quantifying structures is a key issue in biomedical image analysis. The two conventional methods of image segmentation, region-based segmentation and boundary finding, often suffer from a variety of limitations. We propose a method which endeavors to integrate the two approaches in an effort to form a unified approach that is robust to noise and poor initialization. Our approach uses Green's theorem to derive the boundary of a homogeneous region-classified area in the image and integrates this with a grey-level-gradient-based boundary finder. This combines the perceptual notions of edge/shape information with gray level homogeneity.>