Worst-case Model Research Articles

FedCut: A Spectral Analysis Framework for Reliable Detection of Byzantine Colluders.

This paper proposes a general spectral analysis framework that thwarts a security risk in federated Learning caused by groups of malicious Byzantine attackers or colluders, who conspire to upload vicious model updates to severely debase global model performances. The proposed framework delineates the strong consistency and temporal coherence between Byzantine colluders' model updates from a spectral analysis lens, and, formulates the detection of Byzantine misbehaviours as a community detection problem in weighted graphs. The modified normalized graph cut is then utilized to discern attackers from benign participants. Moreover, the Spectral heuristics is adopted to make the detection robust against various attacks. The proposed Byzantine colluder resilient method, i.e., FedCut, is guaranteed to converge with bounded errors. Extensive experimental results under a variety of settings justify the superiority of FedCut, which demonstrates extremely robust model accuracy (MA) under various attacks. It was shown that FedCut's averaged MA is 2.1% to 16.5% better than that of the state of the art Byzantine-resilient methods. In terms of the worst-case model accuracy (MA), FedCut is 17.6% to 69.5% better than these methods.

Load frequency control of uncertain power systems via a novel worst-case GADRC scheme

This article proposes a worst-case generalized active disturbance rejection control (WGADRC) technique for power systems with parametric uncertainties. The proposed scheme is based on the worst-case plant selection strategy. Firstly, a worst-case model of the plant is identified out of the various possible models of an uncertain system, and afterward, the GADRC scheme is employed to design a controller. The best feature of the proposed scheme is that the designed controller ensures superior performance under nominal as well as uncertain plant conditions than other control schemes in the existing literature. To authenticate the supremacy of the proposed methodology, the load frequency control (LFC) issues associated with numerous power systems such as single-area thermal power system, IEEE 39 bus 10 machine test power system, and higher order hydro-thermal power system involving unmodeled parametric uncertainties, unexpected nonlinearities, and physical discrepancies are considered.

Determination of the Worst Case for the Ballistic Test of the Soft Armour System Using the 9mm FMJ Bullets Differing in the Structure

Abstract Ballistic tests require significant rigor and the development of a worst case model during the research processes. The purpose of this study was to evaluate the effect of bullet type (manufacturer) on V50 and Behind Armor Blunt Trauma (BABT) results for two ballistic applications: p-aramid and UHMWPE fibre. The results confirmed the thesis that the source of the bullets implies the test results obtained in terms of the number of penetrated layers in the ballistic system, backface signature deformation profiles (p-BFS) and the level of residual energy transferred to the user of the personal protection.

Buckling optimization of additively manufactured cellular structures using numerical homogenization based on beam models

Interest in components with detailed structures increased with the progress in advanced manufacturing techniques. Parts with lattice elements can provide improved global buckling stability compared to solid structures of the same weight. However, thin features are prone to local buckling. We present a two-scale optimization approach that simultaneously improves the local and global stability of parametrized graded lattice structures. Elastic properties and local buckling behavior are upscaled via homogenization based on geometric exact beam theory. To reduce computational effort, we construct a worst-case model for the homogenized buckling load factor, which acts as a safeguard against local buckling. We briefly discuss advantages and limitations by means of numerical examples.

Orlicz risks for assessing stochastic streamflow environments: a static optimization approach

This study applies novel risk measures, called Orlicz risks, to the risk and uncertainty evaluation of the streamflow discharge as a primary driver of hydrological and hydraulic processes of interest in civil and environmental engineering. We consider the mixed moving average process governing the discharge whose statistics are explicitly represented as some product of a time-scale characterizing the flow attenuation and a jump moment governing the size and frequency of jumps. The classical Orlicz risks are extended so that not only the upper tail risk but also the lower one of the jump size and attenuation of the discharge can be evaluated within a single mathematical framework. Further, the risk and uncertainty can be individually quantified in a tractable manner by the proposed Orlicz risks. Computing the Orlicz risks reduces to solving a pair of novel static optimization problems that are solvable semi-analytically. The risk and uncertainty involved in the streamflow dynamics can be consistently evaluated by specifying few user-dependent parameters. The associated Radon–Nikodym derivatives as the worst-case model uncertainties are obtained as byproducts. Sufficient conditions for the well-posedness of the Orlicz risks are discussed and numerical algorithms for computing them are presented. We finally apply the proposed framework to a statistical analysis of the streamflow discharge time series data collected at mountainous river environments.

Two-scale optimization of graded lattice structures respecting buckling on micro- and macroscale

Interest in components with detailed structures increased with the progress in advanced manufacturing techniques in recent years. Parts with graded lattice elements can provide interesting mechanical, thermal, and acoustic properties compared to parts where only coarse features are included. One of these improvements is better global buckling resistance of the component. However, thin features are prone to local buckling. Normally, analyses with high-computational effort are conducted on high-resolution finite element meshes to optimize parts with good global and local stability. Until recently, works focused only on either global or local buckling behavior. We use two-scale optimization based on asymptotic homogenization of elastic properties and local buckling behavior to reduce the effort of full-scale analyses. For this, we present an approach for concurrent local and global buckling optimization of parameterized graded lattice structures. It is based on a worst-case model for the homogenized buckling load factor, which acts as a safeguard against pure local buckling. Cross-modes residing on both scales are not detected. We support our theory with numerical examples and validations on dehomogenized designs, which show the capabilities of our method, and discuss the advantages and limitations of the worst-case model.

Guaranteeing a Physically Realizable Battery Dispatch Without Charge-Discharge Complementarity Constraints

The non-convex complementarity constraints present a fundamental computational challenge in energy constrained optimization problems. In this work, we present a new, linear, and robust battery optimization formulation that sidesteps the need for battery complementarity constraints and integers and prove analytically that the formulation guarantees that all energy constraints are satisfied which ensures that the optimized battery dispatch is physically realizable. In addition, we bound the worst-case model mismatch and discuss conservativeness. Simulation results further illustrate the effectiveness of this approach.

Data-driven worst case model predictive control algorithm for propylene distillation column under uncertainty of top composition

Composition soft sensors have wide application in the distillation process. In this study, considering the limitation of the prediction ability of the composition soft sensor, a data-driven worst case model predictive control of propylene distillation column is proposed to hedge against the uncertainty of top composition. Firstly, based on the compartmental method and the dynamic mechanism model, a linear state space model of the distillation column is constructed. Aiming at dealing with the uncertainty of top propylene content caused by composition soft sensor, the data-driven uncertainty set is constructed by the combination of principal component analysis and kernel density estimation based on the historical data. Then, the certainty equivalent, traditional worst case, data-driven worst case, set-point tracking and offset-free model predictive control algorithm are designed. Finally, a case study of composition control in a propylene distillation column is carried out. Compared with other strategies, the proposed algorithm ensures the quality of the product while achieving small quality surplus and low operating cost.

Don’t Learn What You Already Know

Over the past few years, deep-learning-based attacks have emerged as a de facto standard, thanks to their ability to break implementations of cryptographic primitives without pre-processing, even against widely used counter-measures such as hiding and masking. However, the recent works of Bronchain and Standaert at Tches 2020 questioned the soundness of such tools if used in an uninformed setting to evaluate implementations protected with higher-order masking. On the opposite, worst-case evaluations may be seen as possibly far from what a real-world adversary could do, thereby leading to too conservative security bounds. In this paper, we propose a new threat model that we name scheme-aware benefiting from a trade-off between uninformed and worst-case models. Our scheme-aware model is closer to a real-world adversary, in the sense that it does not need to have access to the random nonces used by masking during the profiling phase like in a worst-case model, while it does not need to learn the masking scheme as implicitly done by an uninformed adversary. We show how to combine the power of deep learning with the prior knowledge of scheme-aware modeling. As a result, we show on simulations and experiments on public datasets how it sometimes allows to reduce by an order of magnitude the profiling complexity, i.e., the number of profiling traces needed to satisfyingly train a model, compared to a fully uninformed adversary.

Shape suitability and mechanical safety of customised hip implants: Three-dimensional printed acetabular cup for hip arthroplasty

BackgroundOwing to an increase in the number of hip arthroplasty surgeries, the number of implant replacement surgeries is increasing rapidly as well. This necessitates the study of hip joint conditions. Therefore, Paprosky defined a classification system to indicate the degree of damage to the hip joint. In this study, a customised hip implant suitable for Paprosky classification Type ⅡC and over was designed. The shape, suitability, and mechanical safety of the worst-case model for the implant were evaluated. Materials and methodsTo identify the implant size depending on states over Type ⅡC acetabulum bone loss, a size range was selected and a customised implant was designed according to the computed tomography data within the size range. The implant was designed for the flange, hook, and flattened model types. The worst-case selection test was conducted using finite element analysis. The von Mises stresses of the flange, hook, and flattened models were found as 76.223, 136.99, and 80.791 MPa, respectively. Therefore, the hook-type model was selected as the worst case for the mechanical performance test. ResultsA bending test was conducted on the hook-type model without fracture and failure at 5344.56 N. The proposed customised implant was found suitable for Type ⅢA acetabulum bone loss, whereas the shape suitability and mechanical safety were verified for the worst case. ConclusionThe shape of a customised implant suitable for Paprosky ⅢA type was designed. The shape suitability and mechanical safety were evaluated using finite element method analysis and bending tests. Clinical validation is required through subsequent clinical evaluation.

A comparison of approaches to improve worst-case predictive model performance over patient subpopulations

Predictive models for clinical outcomes that are accurate on average in a patient population may underperform drastically for some subpopulations, potentially introducing or reinforcing inequities in care access and quality. Model training approaches that aim to maximize worst-case model performance across subpopulations, such as distributionally robust optimization (DRO), attempt to address this problem without introducing additional harms. We conduct a large-scale empirical study of DRO and several variations of standard learning procedures to identify approaches for model development and selection that consistently improve disaggregated and worst-case performance over subpopulations compared to standard approaches for learning predictive models from electronic health records data. In the course of our evaluation, we introduce an extension to DRO approaches that allows for specification of the metric used to assess worst-case performance. We conduct the analysis for models that predict in-hospital mortality, prolonged length of stay, and 30-day readmission for inpatient admissions, and predict in-hospital mortality using intensive care data. We find that, with relatively few exceptions, no approach performs better, for each patient subpopulation examined, than standard learning procedures using the entire training dataset. These results imply that when it is of interest to improve model performance for patient subpopulations beyond what can be achieved with standard practices, it may be necessary to do so via data collection techniques that increase the effective sample size or reduce the level of noise in the prediction problem.

Revisiting mouse minute virusinactivation by high temperature short time treatment.

In recent years, high temperature short time (HTST) treatment technology has been increasingly adopted for medium treatment to mitigate the potential risk of viral contamination in mammalian cell culture GMP manufacturing facilities. Mouse minute virus (MMV), also called minute virus of mice (MVM), implicated in multiple viral contamination events is commonly used as a relevant model virus to assess the effectiveness of HTST treatment of cell culture media. However, results from different studies vary broadly in inactivation kinetics as well as log reduction factors (LRFs) achieved under given treatment conditions. To determine whether the reported discrepancies stemmed from differences in MMV strains, laboratory-scale HTST devices, medium matrices, and/or experimental designs, we have taken a collaborative approach to systematically assess the effectiveness of HTST treatment for MMV inactivation. This effort was conceptualized based on a media treatment gap analysis conducted by the Consortium on Adventitious Agent Contamination in Biomanufacturing (CAACB) under the MIT Center for Biomedical Innovation (CBI). Specifically, two different MMV strains were used to evaluate the effectiveness of HTST at various treatment conditions with regard to exposure temperature and hold time duration by two independent laboratories within two different companies. To minimize experimental variations, the two sites used the same batches of MMV stocks, the same commercially purchased medium, and the same model of thermocyclers as the laboratory-scale HTST device. The two independent laboratories yielded similar MMV inactivation kinetics and comparable LRF. No significant differences were observed between the two MMV strains evaluated, suggesting that the variations from prior studies were likely due to differences in equipment, medium matrices, or other factors. The data presented here indicate that MMV inactivation by HTST treatment obeys first-order kinetics and can be mathematically modeled using an Arrhenius equation. The model-based extrapolation provides a quantitative estimate of MMV inactivation by the current industry standard HTST condition (102°C for a hold time of 10 s) used for medium treatment. Finally, based on the data from the current study and the industry experience, it is recommended that any alternative virus barrier technologies adopted for medium treatment should provide a clearance of at least 3.0 LRF based on a worst-case model virus to effectively mitigate potential risks of viral contamination.

Model risk in mean-variance portfolio selection: an analytic solution to the worst-case approach

In this paper we consider the worst-case model risk approach described in Glasserman and Xu (Quant Finance 14(1):29–58, 2014). Portfolio selection with model risk can be a challenging operational research problem. In particular, it presents an additional optimisation compared to the classical one. We find the analytical solution for the optimal mean-variance portfolio selection in the worst-case scenario approach and for the special case with the additional constraint of a constant mean vector considered in Glasserman and Xu (Quant Finance 14(1):29–58, 2014). Moreover, we prove in two relevant cases—the minimum-variance case and the symmetric case, i.e. when all assets have the same mean—that the analytical solutions in the alternative model and in the nominal one are equal; we show that this corresponds to the situation when model risk reduces to estimation risk.

Design and Implementation of GSO Based PI Controller for Non Linear Spherical Tank System

From literature point of view process parameter control for non linear process found to be issue in process control industries. The main dilemma lies on managing liquid level in globular tank which is non linear in nature owing to its area disparity in cross section consequent to the change in shape. System identification is made with black box model which is nonlinear and approximated to first order plus dead time model using the worst case model parameter criterion. Group Search Optimization based Proportional Integral controller is implemented to control the non linear system. The real time work is implemented in MATLAB using VMAT-01 module. The servo and regulatory responses of GSO based proportional integral controller intended for different operating region is noted and results are compared with the algorithms such as Big Bang - Big Crunch optimization, Particle swarm optimization and Bacterial Foraging optimization . The most admirable controlling method is determined based on indices such as reduced overshoot, better set point tracking, disturbance rejection , lower performance indices, faster settling time and rise time.

Reducing WCET Overestimations in Multi-Thread Loops with Critical Section Usage

Worst-case execution time (WCET) is an important metric in real-time systems that helps in energy usage modeling and predefined execution time requirement evaluation. While basic timing analysis relies on execution path identification and its length evaluation, multi-thread code with critical section usage brings additional complications and requires analysis of resource-waiting time estimation. In this paper, we solve a problem of worst-case execution time overestimation reduction in situations when multiple threads are executing loops with the same critical section usage in each iteration. The experiment showed the worst-case execution time does not take into account the proportion between computational and critical sections; therefore, we proposed a new worst-case execution time calculation model to reduce the overestimation. The proposed model results prove to reduce the overestimation on average by half in comparison to the theoretical model. Therefore, this leads to more accurate execution time and energy consumption estimation.

Robust Secure Beamforming Design for Cooperative Cognitive Radio Nonorthogonal Multiple Access Networks

By the integration of cooperative cognitive radio (CR) and nonorthogonal multiple access (NOMA), cooperative CR NOMA networks can improve the spectrum efficiency of wireless networks significantly. Due to the openness and exposure of wireless signals, secure communication is an important issue for cooperative CR NOMA networks. In this paper, we investigate the physical layer security design for cooperative CR NOMA networks. Our objective is to achieve maximum secrecy rate of the secondary user by designing optimal beamformers and artificial noise covariance matrix at the multiantenna secondary transmitter under the quality-of-service at the primary user and the transmit power constraint at the secondary transmitter. We consider the practical case that the channel state information (CSI) of the eavesdropper is imperfect, and we model the imperfect CSI by the worst-case model. We show that the robust secrecy rate maximization problem can be transformed to a series of semidefinite programmings based on S-procedure and rank-one relaxation. We also propose an effective method to recover the optimal rank-one solution. Simulations are provided to show the effectiveness of our proposed robust secure algorithm with comparison to the nonrobust secure design and traditional orthogonal multiple access schemes.

Estimating robustness

I estimate and evaluate a model with a representative agent who is concerned that the persistence properties of her baseline model of consumption and inflation are misspecified. Coping with model uncertainty, she discovers a pessimistically biased worst-case model that dictates her behavior. I combine interest rates and aggregate macro series with cross-equation restrictions implied by robust control theory to estimate this worst-case distribution and show that (1) the model's predictions about key features of the yield curve are in line with the data, and (2) the degree of pessimism underlying these findings is plausible. Interpreting the worst-case as the agent's subjective belief, I derive model implied interest rate forecasts and compare them with analogous survey expectations. I find that the model can replicate the dynamics and average level of bias found in the survey.

An efficient triangle mesh slicing algorithm for all topologies in additive manufacturing

To date, slicing algorithms for additive manufacturing is the most effective for favourable triangular mesh topologies; worst-case models, where a large percentage of triangles intersect each slice plane, take significantly longer to slice than a like-for-like file. In larger files, this results in a significant slicing duration, when models are both worst cases and contain more than 100,000 triangles. The research presented here introduces a slicing algorithm which can slice worst-case large models effectively. A new algorithm is implemented utilising an efficient contour construction method, with further adaptations, which make the algorithm suitable for all model topologies. Edge matching, which is an advanced sorting method, decreases the number of sorts per edge from n total number of intersections to two, alongside additional micro-optimisations that deliver the enhanced efficient contour construction algorithm. The algorithm was able to slice a worst-case model of 2.5 million triangles in the 1025s. Maximum improvement was measured as 9400% over the standard efficient contour construction method. Improvements were also observed in all parts in excess of 1000 triangles. The slicing algorithm presented offers novel methods that address the failings of other algorithms described in literature to slice worst-case models effectively.

Mechanistic insights into viral clearance during the chromatography steps in antibody processes by using virus surrogates

Viral safety is required for biological products to treat human diseases, and the burden of inactivation and or virus removal lies on the downstream purification process. Minute virus of mice (MVM) is a nonenveloped parvovirus commonly used as the worst-case model virus in validation studies because of its small size and high chemical stability. In this study, we investigated the use of MVM-mock virus particle (MVP) and bacteriophage ΦX174 as surrogates for MVM to mimic viral clearance studies, with a focus on chromatography operations. Based on structural models and comparison of log reduction value among MVM, MVP, and ΦX174, it was demonstrated that MVP can be used as a noninfectious surrogate to assess viral clearance during process development in multiple chromatography systems in a biosafety level one (BSL-1) laboratory. Protein A (ProA) chromatography was investigated to strategically assess the impact of the resin, impurities, and the monoclonal antibody product on virus removal.

Informationally Robust Optimal Auction Design

A single unit of a good is to be sold by auction to one of two buyers. The good has either a high value or a low value, with known prior probabilities. The designer of the auction knows the prior over values but is uncertain about the correct model of the buyers’ beliefs. The designer evaluates a given auction design by the lowest expected revenue that would be generated across all models of buyers’ information that are consistent with the common prior and across all Bayesian equilibria. An optimal auction for such a seller is constructed, as is a worst-case model of buyers’ information. The theory generates upper bounds on the seller’s optimal payoff for general many-player and common-value models.