Author name cluster

David Garlan

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

7 papers

2 author rows

ECAI Conference 2024 Conference Paper

Error-Driven Uncertainty Aware Training

Pedro Mendes
Paolo Romano 0002
David Garlan

Neural networks are often overconfident about their predictions, which undermines their reliability and trustworthiness. In this work, we present a novel technique, named Error-Driven Uncertainty Aware Training (EUAT), which aims to enhance the ability of neural classifiers to estimate their uncertainty correctly, namely to be highly uncertain when they output inaccurate predictions and low uncertain when their output is accurate. The EUAT approach operates during the model’s training phase by selectively employing two loss functions depending on whether the training examples are correctly or incorrectly predicted by the model. This allows for pursuing the twofold goal of i) minimizing model uncertainty for correctly predicted inputs and ii) maximizing uncertainty for mispredicted inputs, while preserving the model’s misprediction rate. We evaluate EUAT using diverse neural models and datasets in the image recognition domains considering both non-adversarial and adversarial settings. The results show that EUAT outperforms existing approaches for uncertainty estimation (including other uncertainty-aware training techniques, calibration, ensembles, and DEUP) by providing uncertainty estimates that not only have higher quality when evaluated via statistical metrics (e. g. , correlation with residuals) but also when employed to build binary classifiers that decide whether the model’s output can be trusted or not and under distributional data shifts.

Details

TAAS Journal 2024 Journal Article

Self-adapting Machine Learning-based Systems via a Probabilistic Model Checking Framework

Maria Casimiro
Diogo Soares
David Garlan
Luís Rodrigues
Paolo Romano

This article focuses on the problem of optimizing the system utility of Machine Learning (ML)-based systems in the presence of ML mispredictions. This is achieved via the use of self-adaptive systems and through the execution of adaptation tactics, such as model retraining, which operate at the level of individual ML components. To address this problem, we propose a probabilistic modeling framework that reasons about the cost/benefit tradeoffs associated with adapting ML components. The key idea of the proposed approach is to decouple the problems of estimating (1) the expected performance improvement after adaptation and (2) the impact of ML adaptation on overall system utility. We apply the proposed framework to engineer a self-adaptive ML-based fraud detection system, which we evaluate using a publicly available, real fraud detection dataset. We initially consider a scenario in which information on the model’s quality is immediately available. Next, we relax this assumption by integrating (and extending) state-of-the-art techniques for estimating the model’s quality in the proposed framework. We show that by predicting the system utility stemming from retraining an ML component, the probabilistic model checker can generate adaptation strategies that are significantly closer to the optimal, as compared against baselines such as periodic or reactive retraining.

Details DOI

AAAI Conference 2023 Conference Paper

HyperJump: Accelerating HyperBand via Risk Modelling

Pedro Mendes
Maria Casimiro
Paolo Romano
David Garlan

In the literature on hyper-parameter tuning, a number of recent solutions rely on low-fidelity observations (e.g., training with sub-sampled datasets) to identify promising configurations to be tested via high-fidelity observations (e.g., using the full dataset). Among these, HyperBand is arguably one of the most popular solutions, due to its efficiency and theoretically provable robustness. In this work, we introduce HyperJump, a new approach that builds on HyperBand’s robust search strategy and complements it with novel model-based risk analysis techniques that accelerate the search by skipping the evaluation of low risk configurations, i.e., configurations that are likely to be eventually discarded by HyperBand. We evaluate HyperJump on a suite of hyper-parameter optimization problems and show that it provides over one-order of magnitude speed-ups, both in sequential and parallel deployments, on a variety of deep-learning, kernel-based learning and neural architectural search problems when compared to HyperBand and to several state-of-the-art optimizers.

PDF Details DOI

TAAS Journal 2022 Journal Article

Modeling and Analysis of Explanation for Secure Industrial Control Systems

Sridhar Adepu
Nianyu Li
Eunsuk Kang
David Garlan

Many self-adaptive systems benefit from human involvement and oversight, where a human operator can provide expertise not available to the system and detect problems that the system is unaware of. One way of achieving this synergy is by placing the human operator on the loop —i.e., providing supervisory oversight and intervening in the case of questionable adaptation decisions. To make such interaction effective, an explanation can play an important role in allowing the human operator to understand why the system is making certain decisions and improve the level of knowledge that the operator has about the system. This, in turn, may improve the operator’s capability to intervene and, if necessary, override the decisions being made by the system. However, explanations may incur costs, in terms of delay in actions and the possibility that a human may make a bad judgment. Hence, it is not always obvious whether an explanation will improve overall utility and, if so, then what kind of explanation should be provided to the operator. In this work, we define a formal framework for reasoning about explanations of adaptive system behaviors and the conditions under which they are warranted. Specifically, we characterize explanations in terms of explanation content, effect, and cost. We then present a dynamic system adaptation approach that leverages a probabilistic reasoning technique to determine when an explanation should be used to improve overall system utility. We evaluate our explanation framework in the context of a realistic industrial control system with adaptive behaviors.

Details DOI

TAAS Journal 2020 Journal Article

Information Reuse and Stochastic Search

Cody Kinneer
David Garlan
Claire Le Goues

Many software systems operate in environments of change and uncertainty. Techniques for self-adaptation allow these systems to automatically respond to environmental changes, yet they do not handle changes to the adaptive system itself, such as the addition or removal of adaptation tactics. Instead, changes in a self-adaptive system often require a human planner to redo an expensive planning process to allow the system to continue satisfying its quality requirements under different conditions; automated techniques must replan from scratch. We propose to address this problem by reusing prior planning knowledge to adapt to unexpected situations. We present a planner based on genetic programming that reuses existing plans and evaluate this planner on two case-study systems: a cloud-based web server and a team of autonomous aircraft. While reusing material in genetic algorithms has been recently applied successfully in the area of automated program repair, we find that naively reusing existing plans for self- * planning can actually result in a utility loss. Furthermore, we propose a series of techniques to lower the costs of reuse, allowing genetic techniques to leverage existing information to improve utility when replanning for unexpected changes, and we find that coarsely shaped search-spaces present profitable opportunities for reuse.

Details DOI

TAAS Journal 2018 Journal Article

Flexible and Efficient Decision-Making for Proactive Latency-Aware Self-Adaptation

Gabriel A. Moreno
Javier Cámara
David Garlan
Bradley Schmerl

Proactive latency-aware adaptation is an approach for self-adaptive systems that considers both the current and anticipated adaptation needs when making adaptation decisions, taking into account the latency of the available adaptation tactics. Since this is a problem of selecting adaptation actions in the context of the probabilistic behavior of the environment, Markov decision processes (MDPs) are a suitable approach. However, given all the possible interactions between the different and possibly concurrent adaptation tactics, the system, and the environment, constructing the MDP is a complex task. Probabilistic model checking has been used to deal with this problem, but it requires constructing the MDP every time an adaptation decision is made to incorporate the latest predictions of the environment behavior. In this article, we describe PLA-SDP, an approach that eliminates that runtime overhead by constructing most of the MDP offline. At runtime, the adaptation decision is made by solving the MDP through stochastic dynamic programming, weaving in the environment model as the solution is computed. We also present extensions that support different notions of utility, such as maximizing reward gain subject to the satisfaction of a probabilistic constraint, making PLA-SDP applicable to systems with different kinds of adaptation goals.

Details DOI

TAAS Journal 2016 Journal Article

Analyzing Latency-Aware Self-Adaptation Using Stochastic Games and Simulations

Javier Cámara
Gabriel A. Moreno
David Garlan
Bradley Schmerl

Self-adaptive systems must decide which adaptations to apply and when. In reactive approaches, adaptations are chosen and executed after some issue in the system has been detected (e.g., unforeseen attacks or failures). In proactive approaches, predictions are used to prepare the system for some future event (e.g., traffic spikes during holidays). In both cases, the choice of adaptation is based on the estimated impact it will have on the system. Current decision-making approaches assume that the impact will be instantaneous, whereas it is common that adaptations take time to produce their impact. Ignoring this latency is problematic because adaptations may not achieve their effect in time for a predicted event. Furthermore, lower impact but quicker adaptations may be ignored altogether, even if over time the accrued impact is actually higher. In this article, we introduce a novel approach to choosing adaptations that considers these latencies. To show how this improves adaptation decisions, we use a two-pronged approach: (i) model checking of Stochastic Multiplayer Games (SMGs) enables us to understand best- and worst-case scenarios of optimal latency-aware and non-latency-aware adaptation without the need to develop specific adaptation algorithms. However, since SMGs do not provide an algorithm to make choices at runtime, we propose a (ii) latency-aware adaptation algorithm to make decisions at runtime. Simulations are used to explore more detailed adaptation behavior and to check if the performance of the algorithm falls within the bounds predicted by SMGs. Our results show that latency awareness improves adaptation outcomes and also allows a larger set of adaptations to be exploited.

Details DOI