Tanmay Khandait

In this paper we propose for the first time the hyperparameter optimization (HPO) algorithm POCAII. POCAII differs from the Hyperband and Successive Halving literature by explicitly separating the search and evaluation phases and utilizing principled approaches to exploration and exploitation principles during both phases. Such distinction results in a highly flexible scheme for managing a hyperparameter optimization budget by focusing on search (i.e., generating competing configurations) towards the start of the HPO process while increasing the evaluation effort as the HPO comes to an end. POCAII was compared to state of the art approaches SMAC, BOHB and DEHB. Our algorithm shows superior performance in low-budget hyperparameter optimization regimes. Since many practitioners do not have exhaustive resources to assign to HPO, it has wide applications to real-world problems. Moreover, the empirical evidence showed how POCAII demonstrates higher robustness and lower variance in the results. This is again very important when considering realistic scenarios with extremely expensive models to train.

The performance of modern machine learning algorithms depends upon the selection of a set of hyperparameters. Common examples of hyperparameters are learning rate and the number of layers in a dense neural network. Auto-ML is a branch of optimization that has produced important contributions in this area. Within Auto-ML, hyperband-based approaches, which eliminate poorly-performing configurations after evaluating them at low budgets, are among the most effective. However, the performance of these algorithms strongly depends on how effectively they allocate the computational budget to various hyperparameter configurations. We present the new Parameter Optimization with Conscious Allocation (POCA), a hyperband-based algorithm that adaptively allocates the inputted budget to the hyperparameter configurations it generates following a Bayesian sampling scheme. We compare POCA to its nearest competitor at optimizing the hyperparameters of an artificial toy function and a deep neural network and find that POCA finds strong configurations faster in both settings.

Requirements driven search-based testing (also known as falsification) has proven to be a practical and effective method for discovering erroneous behaviors in Cyber-Physical Systems. Despite the constant improvements on the performance and applicability of falsification methods, they all share a common characteristic. Namely, they are best-effort methods which do not provide any guarantees on the absence of erroneous behaviors (falsifiers) when the testing budget is exhausted. The absence of finite time guarantees is a major limitation which prevents falsification methods from being utilized in certification procedures. In this paper, we address the finite-time guarantees problem by developing a new stochastic algorithm. Our proposed algorithm not only estimates (bounds) the probability that falsifying behaviors exist, but also identifies the regions where these falsifying behaviors may occur. We demonstrate the applicability of our approach on standard benchmark functions from the optimization literature and on the F16 benchmark problem. Note to Practitioners —The safety assurance problem for Cyber-Physical Systems (CPS) remains an open challenge. To demonstrate functional safety, practitioners must collect evidence that establishes that a system performs as expected under certain assumptions. The expected system behavior is typically captured through functional correctness requirements. In the case of CPS, evidence typically takes the form of test cases that are executed both on a model of the system and/or on the actual system. One of the challenges in producing such evidence is how to automatically generate test cases which are representative of the infinite execution space of CPS. Search-based test generation (SBTG) is a class of methods that can automatically generate test cases for CPS while being guided by the functional requirements. As SBTG methods try to discover test cases that invalidate, i.e., falsify, the requirements, they also collect validating, i.e., satisfying, test cases that can be used as evidence. This work introduces a method that can assess whether enough test cases have been executed given a finite testing budget. The sufficiency of the test suite is assessed by computing the probability that invalidating system behaviors may exist but have not yet been discovered. The practitioner can then adjust the number of test cases generated until a desired degree of confidence on the probability is achieved. Hence, our method not only works as an automated test case generation algorithm, but also as a method that provides formal functional performance guarantees on the system. Future directions will investigate extensions of our method to stochastic CPS.

We propose a framework for security vulnerability analysis for Cyber-Physical Systems (CPS). Our framework imposes only minimal assumptions on the structure of the CPS. Namely, we consider CPS with feedback control loops, state observers, and anomaly detection algorithms. Moreover, our framework does not require any knowledge about the dynamics or the algorithms used in the CPS. Under this common CPS architecture, we develop tools that can identify vulnerabilities in the system and their impact on the functionality of the CPS. We pose the CPS security problem as a falsification (or Search Based Test Generation (SBTG)) problem guided by security requirements expressed in Signal Temporal Logic (STL). We propose two different categories of security requirements encoded in STL: (1) detectability (stealthiness) and (2) effectiveness (impact on the CPS function). Finally, we demonstrate in simulation on an inverted pendulum and on an Unmanned Aerial Vehicle (UAV) that both specifications are falsifiable using our SBTG techniques.

In the paper titled " Stealthy attacks formalized as STL formulas for Falsification of CPS Security", we investigate a broad class of attacks on the sensor and actuation blocks in the form of additive perturbation that impacts the measurement and control, respectively. In this demo, we demonstrate the usage of our framework and the underlying technologies along with a case study on aviation systems using Microsoft Flight Simulator (MSFS).

Hyperparameter tuning is an active area of research in machine learning, where the aim is to identify the optimal hyperparameters that provide the best performance on the validation set. Hyperparameter tuning is often achieved using naive techniques, such as random search and grid search. However, most of these methods seldom lead to an optimal set of hyperparameters and often get very expensive. In this paper, we propose a bilevel solution method for solving the hyperparameter optimization problem that does not suffer from the drawbacks of the earlier studies. The proposed method is general and can be easily applied to any class of machine learning algorithms. The idea is based on the approximation of the lower level optimal value function mapping, which is an important mapping in bilevel optimization and helps in reducing the bilevel problem to a single level constrained optimization task. The single-level constrained optimization problem is solved using the augmented Lagrangian method. We discuss the theory behind the proposed algorithm and perform extensive computational study on two datasets that confirm the efficiency of the proposed method. We perform a comparative study against grid search, random search and Bayesian optimization techniques that shows that the proposed algorithm is multiple times faster on problems with one or two hyperparameters. The computational gain is expected to be significantly higher as the number of hyperparameters increase. Corresponding to a given hyperparameter most of the techniques in the literature often assume a unique optimal parameter set that minimizes loss on the training set. Such an assumption is often violated by deep learning architectures and the proposed method does not require any such assumption.