Maximum Information Gain Research Articles

Digital Twin Concepts with Uncertainty for Nuclear Power Applications

Digital Twins (DTs) are receiving considerable attention from multiple disciplines. Much of the literature at this time is dedicated to the conceptualization of digital twins, and associated enabling technologies and challenges. In this paper, we consider these propositions for the specific application of nuclear power. Our review finds that the current DT concepts are amenable to nuclear power systems, but benefit from some modifications and enhancements. Further, some areas of the existing modeling and simulation infrastructure around nuclear power systems are adaptable to DT development, while more recent efforts in advanced modeling and simulation are less suitable at this time. For nuclear power applications, DT development should rely first on mechanistic model-based methods to leverage the extensive experience and understanding of these systems. Model-free techniques can then be adopted to selectively, and correctively, augment limitations in the model-based approaches. Challenges to the realization of a DT are also discussed, with some being unique to nuclear engineering, however most are broader. A challenging aspect we discuss in detail for DTs is the incorporation of uncertainty quantification (UQ). Forward UQ enables the propagation of uncertainty from the digital representations to predict behavior of the physical asset. Similarly, inverse UQ allows for the incorporation of data from new measurements obtained from the physical asset back into the DT. Optimization under uncertainty facilitates decision support through the formal methods of optimal experimental design and design optimization that maximize information gain, or performance, of the physical asset in an uncertain environment.

Determining the maximum information gain and optimizing experimental design in neutron reflectometry using the Fisher information.

An approach based on the Fisher information (FI) is developed to quantify the maximum information gain and optimal experimental design in neutron reflectometry experiments. In these experiments, the FI can be calculated analytically and used to provide sub-second predictions of parameter uncertainties. This approach can be used to influence real-time decisions about measurement angle, measurement time, contrast choice and other experimental conditions based on parameters of interest. The FI provides a lower bound on parameter estimation uncertainties, and these are shown to decrease with the square root of the measurement time, providing useful information for the planning and scheduling of experimental work. As the FI is computationally inexpensive to calculate, it can be computed repeatedly during the course of an experiment, saving costly beam time by signalling that sufficient data have been obtained or saving experimental data sets by signalling that an experiment needs to continue. The approach's predictions are validated through the introduction of an experiment simulation framework that incorporates instrument-specific incident flux profiles, and through the investigation of measuring the structural properties of a phospholipid bilayer.

Strategic testing approaches for targeted disease monitoring can be used to inform pandemic decision-making.

More than 1.6 million Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) tests were administered daily in the United States at the peak of the epidemic, with a significant focus on individual treatment. Here, we show that objective-driven, strategic sampling designs and analyses can maximize information gain at the population level, which is necessary to increase situational awareness and predict, prepare for, and respond to a pandemic, while also continuing to inform individual treatment. By focusing on specific objectives such as individual treatment or disease prediction and control (e.g., via the collection of population-level statistics to inform lockdown measures or vaccine rollout) and drawing from the literature on capture–recapture methods to deal with nonrandom sampling and testing errors, we illustrate how public health objectives can be achieved even with limited test availability when testing programs are designed a priori to meet those objectives.

An innovative approach to identify environmental variables with conservation priorities in habitat patches

Conservation of habitat patches and the related environment benefits both the focal species and human well-being. Many indices use the dispersal range to identify habitat patches with conservation priorities. However, there lacks approaches to identify environmental variables with conservation priorities (noted as target variables) in those identified patches. Therefore, this paper proposes an approach to identify environmental variables with conservation priorities in habitat patches using perception range and introduces the related assumption. It is assumed the agents select habitats based on their prior preference and perceived information in their perception ranges, which avoids the omniscient assumption of agents. Based on such assumptions, the proposed approach identifies the target variables by approximating how animals identify their habitats. It highlights the use of perception range and identifies target variables using the maximum information gain. The variables that contribute the largest reduction of uncertainty are regarded as the target variables in the habitat patches. Taking the Common Moorhen (Gallinula chloropus) living in Tianjin, China as the case, different scenarios with 100 m, 250 m and 500 m perception ranges are designed to illustrate the feasibility of the proposed approach. The proposed approach identifies the normalized vegetation index, rather than the distance to water surface, is the target variable in 42.3%, 58.9% and 72.1% habitat patches with given perception ranges. Adjustments are made on areas within the given perception range of each patch. More grid cells that has increased suitability index can be found in scenarios given 250 m perception range, which indicates the conservation area is not always the large the better. Optimizations are expected on both a better approximation method and a more thorough hypothesis of using perception range.

Estimating α-Rank by Maximizing Information Gain

Game theory has been increasingly applied in settings where the game is not known outright, but has to be estimated by sampling. For example, meta-games that arise in multi-agent evaluation can only be accessed by running a succession of expensive experiments that may involve simultaneous deployment of several agents. In this paper, we focus on α-rank, a popular game-theoretic solution concept designed to perform well in such scenarios. We aim to estimate the α-rank of the game using as few samples as possible. Our algorithm maximizes information gain between an epistemic belief over the α-ranks and the observed payoff. This approach has two main benefits. First, it allows us to focus our sampling on the entries that matter the most for identifying the α-rank. Second, the Bayesian formulation provides a facility to build in modeling assumptions by using a prior over game payoffs. We show the benefits of using information gain as compared to the confidence interval criterion of ResponseGraphUCB, and provide theoretical results justifying our method.

Risk-aware sensor scheduling and tracking of large constellations

Commercial proposals for large constellations of satellites present a major problem in the field of Space Situational Awareness (SSA). Space traffic management systems must produce accurate and timely predictions of potential collisions for operators to perform evasive maneuvers. This requires precise orbit determination solutions, but limitations on sensor resources can impair tracker performance. This work presents an integrated tasking and tracking system that can autonomously task a network of ground sensors to track large constellations with the goal of identifying hazardous conjunctions. The sensor scheduling approach combines information gain maximization with predicted probability of collision to produce a single reward function for tasking. The labeled multi-Bernoulli filter, a multi-target filter capable of resolving data ambiguity, processes the data and the resulting catalog estimate drives another round of scheduling. A tracking simulation involving 4,545 satellites in low-Earth orbit and a network of twelve ground-based sensors demonstrates how this scheduling method can focus on high-risk targets while maintaining custody of the entire constellation. A conjunction assessment is performed using the estimated catalog at the end of the tracking phase, and the accuracy of the analysis is used to evaluate the scheduling method’s efficacy.

Optimizing Non-Markovian Information Gain Under Physics-Based Communication Constraints

In many exploration scenarios, it is important for robots to efficiently explore new areas and constantly communicate results. Mobile robots inherently couple motion and network topology due to the effects of position on wireless propagation-e.g. distance or obstacles between network nodes. Information gain is a useful measure of exploration. However, finding paths that maximize information gain while preserving communication is challenging due to the non-Markovian nature of information gain, discontinuities in network topology, and zero-reward local optima. We address these challenges through an optimization and sampling-based algorithm. Our algorithm scales to 50% more robots and obtains 2-5 times more information relative to path cost compared to baseline planning approaches.

Maximum information gain of approximate quantum position measurement

We perform a quantum information analysis for multi-mode Gaussian approximate position measurements, underlying noisy homodyning in quantum optics. The “Gaussian maximizer” property is established for the entropy reduction of these measurements which provides explicit formulas for computations of their maximum information gain or entanglement-assisted capacity. The case of one mode is discussed in detail.

Recycling the resource: Sequential usage of shared state in quantum teleportation with weak measurements

Complete measurements, while providing maximal information gain, result in destruction of the shared entanglement. In the standard teleportation scheme, the sender's measurement on the shared entangled state between the sender and the receiver has that consequence. We propose here a teleportation scheme involving weak measurements which can sustain entanglement up to a certain level so that the reusability of the shared resource state for reattempting teleportation is possible. The measurements are chosen in such a way that it is weak enough to retain entanglement and hence can be reused for quantum tasks, yet adequately strong to ensure quantum advantage in the protocol. In this scenario, we report that at most six sender-receiver duos can reuse the state for reattempting teleportation, when the initial shared state is entangled in a finite neighborhood of the maximally entangled state and for a suitable choice of weak measurements. However, we observe that the reattempt number decreases with the decrease in the entanglement of the initial shared state. Among the weakening strategies studied, Bell measurement admixed with white noise performs better than any other low-rank weak measurements in this situation.

Safe Optimal Design with Applications in Policy Learning

Motivated by practical needs in online experimentation and off-policy learning, we study the problem of safe optimal design, where we develop a data logging policy that efficiently explores while achieving competitive rewards with a baseline production policy. We first show, perhaps surprisingly, that a common practice of mixing the production policy with uniform exploration, despite being safe, is sub-optimal in maximizing information gain. Then we propose a safe optimal logging policy for the case when no side information about the actions’ expected rewards is available. We improve upon this design by considering side information and also extend both approaches to a large number of actions with a linear reward model. We analyze how our data logging policies impact errors in off-policy learning. Finally, we empirically validate the benefit of our designs by conducting extensive experiments.

Bayesian Information-Theoretic Calibration of Radiotherapy Sensitivity Parameters for Informing Effective Scanning Protocols in Cancer

With new advancements in technology, it is now possible to collect data for a variety of different metrics describing tumor growth, including tumor volume, composition, and vascularity, among others. For any proposed model of tumor growth and treatment, we observe large variability among individual patients’ parameter values, particularly those relating to treatment response; thus, exploiting the use of these various metrics for model calibration can be helpful to infer such patient-specific parameters both accurately and early, so that treatment protocols can be adjusted mid-course for maximum efficacy. However, taking measurements can be costly and invasive, limiting clinicians to a sparse collection schedule. As such, the determination of optimal times and metrics for which to collect data in order to best inform proper treatment protocols could be of great assistance to clinicians. In this investigation, we employ a Bayesian information-theoretic calibration protocol for experimental design in order to identify the optimal times at which to collect data for informing treatment parameters. Within this procedure, data collection times are chosen sequentially to maximize the reduction in parameter uncertainty with each added measurement, ensuring that a budget of n high-fidelity experimental measurements results in maximum information gain about the low-fidelity model parameter values. In addition to investigating the optimal temporal pattern for data collection, we also develop a framework for deciding which metrics should be utilized at each data collection point. We illustrate this framework with a variety of toy examples, each utilizing a radiotherapy treatment regimen. For each scenario, we analyze the dependence of the predictive power of the low-fidelity model upon the measurement budget.

CheKiPEUQ Intro 2: Harnessing Uncertainties from Data Sets, Bayesian Design of Experiments in Chemical Kinetics**

AbstractWhen choosing experimental conditions, Bayesian statistical tools can predict the experimental choices which will yield the highest information gain. Experimental choices could be temperature, pressure, reaction time, number of measurements, reactor volume, etc.. Three example analyses are presented here, each using the software Chemical Kinetics Parameter Estimation and Uncertainty Quantification (CheKiPEUQ). Information gain is a measure of reduction of uncertainty in a model's parameters. The three chemical system examples presented each illustrate Bayesian Design of Experiments using information gain. In the first chemical example, temperature selection impacts the information gain for the free energy of reaction in a two‐component equilibrium reaction. In the second example, temperature and pressure are explored for a competitive adsorption Langmuir replacement reaction system. The third example is a catalytic membrane reactor which is a culmination of the previous examples. The catalytic membrane reactor has a complex and nonlinear response in the observables which is solved by numerical evaluation. In the three examples, the experimental conditions are treated as design variables for maximizing information gain.

Assessing thermal sensitivity using transient heat and cold stimuli combined with a Bayesian adaptive method in a clinical setting: A proof of concept study.

Quantitative sensory testing of thermal detection abilities is used as a clinical tool to assess the function of pain pathways. The most common procedure to assess thermal sensitivity, the 'method of limits', provides a quick but rough estimate of detection thresholds. Here, we investigate the potential of evaluating not only the threshold but also the slope of the psychometric functions for cold and warm detection. A convenience sample of 15 patients with diabetes mellitus (DM) and 15 age-matched healthy controls (HC) was tested. Thirty brief (100ms) stimuli of each modality were applied to the volar wrist and foot dorsum. Cold and warm stimuli were delivered with a Peltier thermode and a temperature-controlled CO2 laser, respectively. Stimulus intensities were dynamically selected using an adaptive Bayesian algorithm (psi method) maximizing information gain for threshold and slope estimation. ROC analyses were used to assess the ability of slopes, thresholds and the combination of both to discriminate between groups. Assessment of the slope and threshold of the psychometric function for thermal detection took about 10min. The ability to detect warmth was not reduced in DM patients as compared to HC. Cold detection performance assessed using slope or threshold parameters separated DM from HC with good discriminative power. Discrimination was further increased when both parameters were used together (93% sensitivity and 87% specificity), indicating that they provide complementary information on patient status. The psi method may be an interesting alternative to the classical method of limits for thermal QST. Current QST protocols provide an incomplete and potentially biased estimate of sensory detection performance. We propose a method that estimates the slope and the threshold of the psychometric function, defining heat and cold sensory detection performance, in only a few minutes. Furthermore, we provide preliminary evidence that combining slope and threshold parameters of cold detection performance leads to a better discriminative ability than relying solely on the threshold.

A kelly criterion based optimal scheduling of a microgrid on a steam-assisted gravity drainage (SAGD) facility

Steam Assisted Gravity Drainage (SAGD) is one of the major thermal recovery techniques to extract the Canadian bitumen. In the process, there is substantial amount of heat left behind in the reservoir at the end of SAGD operations, which can be converted to electricity and distributed profitably utilizing a microgrid. Concurrently, based on Canada’s green energy initiative, we present the integration and optimal power management of a microgrid on a SAGD facility, comprising of an Organic Rankine Cycle based turbine (ORC), a Gas Turbine (GT), a Battery Storage System (BSS), the central grid and the SAGD facility as the load. We introduce a Kelly Criterion (KC) based microgrid scheduling technique, which is based on maximizing information gain and is independent of supply-demand relationships. Based on the formulation, it was hypothesized that the KC based algorithm is superior compared to standard optimization techniques in a highly volatile electricity market. The electricity market volatility is captured via a wavelet network based forecasting technique. The case study presented highlights the advantages of the KC approach and its efficacy in a highly volatile energy market. The results show the superiority of the KC based scheduling algorithm in highly volatile energy markets, with forecast data irregularities.

Data-Efficient Autotuning With Bayesian Optimization: An Industrial Control Study

Bayesian optimization is proposed for automatic learning of optimal controller parameters from experimental data. A probabilistic description (a Gaussian process) is used to model the unknown function from controller parameters to a user-defined cost. The probabilistic model is updated with data, which is obtained by testing a set of parameters on the physical system and evaluating the cost. In order to learn fast, the Bayesian optimization algorithm selects the next parameters to evaluate in a systematic way, for example, by maximizing information gain about the optimum. The algorithm thus iteratively finds the globally optimal parameters with only few experiments. Taking throttle valve control as a representative industrial control example, the proposed auto-tuning method is shown to outperform manual calibration: it consistently achieves better performance with a low number of experiments. The proposed auto-tuning framework is flexible and can handle different control structures and objectives.

All-relevant feature selection using multidimensional filters with exhaustive search

This paper describes a method for the identification of informative variables in an information system with discrete decision variables. It is targeted specifically towards the discovery of variables that are non-informative when considered alone, but informative when the synergistic interactions between multiple variables are considered. To this end, mutual entropy of all possible k-tuples of variables with a decision variable is computed. Then, for each variable, the maximum information gain due to interactions with other variables is obtained. For non-informative variables, this quantity conforms to well-known statistical distributions. This allows for discerning the truly informative variables from the non-informative ones. To demonstrate this approach, the method is applied to several synthetic datasets that involve complex multidimensional interactions between variables. The performance of the method is also validated on a real-world dataset. It is shown that the method is capable of identifying the most important informative variables, even when the dimensionality of the analysis is smaller than the true dimensionality of the problem. What is more, the high sensitivity of the algorithm allows for the detection of the influence of nuisance variables on the response variable.

Energy consumption and cooperation for optimal sensing

The reliable detection of environmental molecules in the presence of noise is an important cellular function, yet the underlying computational mechanisms are not well understood. We introduce a model of two interacting sensors which allows for the principled exploration of signal statistics, cooperation strategies and the role of energy consumption in optimal sensing, quantified through the mutual information between the signal and the sensors. Here we report that in general the optimal sensing strategy depends both on the noise level and the statistics of the signals. For joint, correlated signals, energy consuming (nonequilibrium), asymmetric couplings result in maximum information gain in the low-noise, high-signal-correlation limit. Surprisingly we also find that energy consumption is not always required for optimal sensing. We generalise our model to incorporate time integration of the sensor state by a population of readout molecules, and demonstrate that sensor interaction and energy consumption remain important for optimal sensing.

Object shape estimation and modeling, based on sparse Gaussian process implicit surfaces, combining visual data and tactile exploration

Inferring and representing three-dimensional shapes is an important part of robotic perception. However, it is challenging to build accurate models of novel objects based on real sensory data, because observed data is typically incomplete and noisy. Furthermore, imperfect sensory data suggests that uncertainty about shapes should be explicitly modeled during shape estimation. Such uncertainty models can usefully enable exploratory action planning for maximum information gain and efficient use of data. This paper presents a probabilistic approach for acquiring object models, based on visual and tactile data. We study Gaussian Process Implicit Surface (GPIS) representation. GPIS enables a non-parametric probabilistic reconstruction of object surfaces from 3D data points, while also providing a principled approach to encode the uncertainty associated with each region of the reconstruction. We investigate different configurations for GPIS, and interpret an object surface as the level-set of an underlying sparse GP. Experiments are performed on both synthetic data, and also real data sets obtained from two different robots physically interacting with objects. We evaluate performance by assessing how close the reconstructed surfaces are to ground-truth object models. We also evaluate how well objects from different categories are clustered, based on the reconstructed surface shapes. Results show that sparse GPs enable a reliable approximation to the full GP solution, and the proposed method yields adequate surface representations to distinguish objects. Additionally the presented approach is shown to provide computational efficiency, and also efficient use of the robot’s exploratory actions.

Incremental Attack Synthesis

Information leakage is a signi cant problem in modern software systems. Information leaks due to side channels are especially hard to detect and analyze. In recent years, techniques have been developed for automated synthesis of adaptive side-channel attacks that recover secret values by iteratively generating inputs to reveal partial information about the secret based on the sidechannel observations. Prominent approaches of attack synthesis use symbolic execution, model counting, and meta-heuristics to maximize information gain. These approaches could bene t by reusing results from prior steps in each step. In this paper, we present an incremental approach to attack synthesis that reuses model counting results from prior iterations in each attack step to improve efficiency. Experimental evaluation demonstrates that our approach drastically improves performance, reducing the attack synthesis time by an order of magnitude.

The multivariate adaptive design for efficient estimation of the time course of perceptual adaptation.

In experiments on behavioral adaptation, hundreds or even thousands of trials per subject are often required in order to accurately recover the many psychometric functions that characterize adaptation's time course. More efficient methods for measuring perceptual changes over time would be beneficial to such efforts. In this article, we propose two methods to adaptively select the optimal stimuli sequentially in an experiment on adaptation: These are the minimum entropy (ME) method and the match probability (MP) method. The ME method minimizes the uncertainty about the joint posterior distribution of the function parameters at each trial and is mathematically equivalent to Zhao, Lesmes, and Lu's (2019) method, which efficiently measures time courses of perceptual change by maximizing information gain. The MP method selects the next stimulus that makes the value of the psychometric function closest to .5-that is, where the probability of choosing either one of the two options for each stimulus is closest to .5. We extended Zhao et al.'s (2019) work by evaluating the ME method in a new domain (contrast adaptation) with two simulation studies that compared it to MP and two other methods (i.e., traditional staircase and random methods), and also explored the optimal block length. ME outperformed the other three methods in general, and using fewer longer blocks generally produced better parameter recovery than using more shorter blocks.