Author name cluster

Joseph Scott

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.

4 papers

2 author rows

AAAI Conference 2024 Short Paper

BertRLFuzzer: A BERT and Reinforcement Learning Based Fuzzer (Student Abstract)

Piyush Jha
Joseph Scott
Jaya Sriram Ganeshna
Mudit Singh
Vijay Ganesh

We present a novel tool BertRLFuzzer, a BERT and Reinforcement Learning (RL) based fuzzer aimed at finding security vulnerabilities for Web applications. BertRLFuzzer works as follows: given a set of seed inputs, the fuzzer performs grammar-adhering and attack-provoking mutation operations on them to generate candidate attack vectors. The key insight of BertRLFuzzer is the use of RL with a BERT model as an agent to guide the fuzzer to efficiently learn grammar-adhering and attack-provoking mutation operators. In order to establish the efficacy of BertRLFuzzer we compare it against a total of 13 black box and white box fuzzers over a benchmark of 9 victim websites with over 16K LOC. We observed a significant improvement, relative to the nearest competing tool in terms of time to first attack (54% less), new vulnerabilities found (17 new vulnerabilities), and attack rate (4.4% more attack vectors generated).

PDF Details DOI

AAAI Conference 2021 Short Paper

Logic Guided Genetic Algorithms (Student Abstract)

Dhananjay Ashok
Joseph Scott
Sebastian J. Wetzel
Maysum Panju
Vijay Ganesh

We present a novel Auxiliary Truth enhanced Genetic Algorithm (GA) that uses logical or mathematical constraints as a means of data augmentation as well as to compute loss with the aim of increasing both data efficiency and accuracy of symbolic regression (SR) algorithms. Our method, logicguided genetic algorithm (LGGA), takes as input a set of labelled datapoints and auxiliary truths (AT) (mathematical facts known a priori about the unknown function the regressor aims to learn) and outputs a specially generated and curated dataset that can be used with any SR method. We evaluate LGGA against state-of-the-art SR tools, namely, Eureqa and TuringBot, and find that using these SR tools in conjunction with LGGA results in them solving up to 30. 0% more equations, needing only a fraction of the amount of data compared to the same tool without LGGA, i. e. , resulting in up to a 61. 9% improvement in data efficiency.

PDF Details

AAAI Conference 2020 Short Paper

LGML: Logic Guided Machine Learning (Student Abstract)

Joseph Scott
Maysum Panju
Vijay Ganesh

We introduce Logic Guided Machine Learning (LGML), a novel approach that symbiotically combines machine learning (ML) and logic solvers to learn mathematical functions from data. LGML consists of two phases, namely a learning-phase and a logic-phase with a corrective feedback loop, such that, the learning-phase learns symbolic expressions from input data, and the logic-phase cross verifies the consistency of the learned expression with known auxiliary truths. If inconsistent, the logic-phase feeds back "counterexamples" to the learning-phase. This process is repeated until the learned expression is consistent with auxiliary truth. Using LGML, we were able to learn expressions that correspond to the Pythagorean theorem and the sine function, with several orders of magnitude improvements in data efficiency compared to an approach based on an out-of-the-box multi-layered perceptron (MLP).

PDF Details

SAT Conference 2017 Conference Paper

A Propagation Rate Based Splitting Heuristic for Divide-and-Conquer Solvers

Saeed Nejati
Zack Newsham
Joseph Scott
Jia Hui Liang
Catherine H. Gebotys
Pascal Poupart
Vijay Ganesh 0001

Abstract In this paper, we present a divide-and-conquer SAT solver, M apleAmpharos, that uses a novel propagation-rate (PR) based splitting heuristic. The key idea is that we rank variables based on the ratio of how many propagations they cause during the run of the worker conflict-driven clause-learning solvers to the number of times they are branched on, with the variable that causes the most propagations ranked first. The intuition here is that, in the context of divide-and-conquer solvers, it is most profitable to split on variables that maximize the propagation rate. Our implementation M apleAmpharos uses the AMPHAROS solver as its base. We performed extensive evaluation of M apleAmpharos against other competitive parallel solvers such as Treengeling, Plingeling, Parallel CryptoMiniSat5, and Glucose-Syrup. We show that on the SAT 2016 competition Application benchmark and a set of cryptographic instances, our solver M apleAmpharos is competitive with respect to these top parallel solvers. What is surprising that we obtain this result primarily by modifying the splitting heuristic.

Details