Elijah Pelofske

Elijah Pelofske, Vincent Urias, Lorie M. Liebrock

Generative Pre-Trained Transformer models have been shown to be surprisingly effective at a variety of natural language processing tasks -- including generating computer code. We evaluate the effectiveness of open source GPT models for the task of automatic identification of the presence of vulnerable code syntax (specifically targeting C and C++ source code). This task is evaluated on a selection of 36 source code examples from the NIST SARD dataset, which are specifically curated to not contain natural English that indicates the presence, or lack thereof, of a particular vulnerability. The NIST SARD source code dataset contains identified vulnerable lines of source code that are examples of one out of the 839 distinct Common Weakness Enumerations (CWE), allowing for exact quantification of the GPT output classification error rate. A total of 5 GPT models are evaluated, using 10 different inference temperatures and 100 repetitions at each setting, resulting in 5,000 GPT queries per vulnerable source code analyzed. Ultimately, we find that the GPT models that we evaluated are not suitable for fully automated vulnerability scanning because the false positive and false negative rates are too high to likely be useful in practice. However, we do find that the GPT models perform surprisingly well at automated vulnerability detection for some of the test cases, in particular surpassing random sampling, and being able to identify the exact lines of code that are vulnerable albeit at a low success rate. The best performing GPT model result found was Llama-2-70b-chat-hf with inference temperature of 0.1 applied to NIST SARD test case 149165 (which is an example of a buffer overflow vulnerability), which had a binary classification recall score of 1.0 and a precision of 1.0 for correctly and uniquely identifying the vulnerable line of code and the correct CWE number.

Kyle Henke, Elijah Pelofske, Georg Hahn et al.

We consider the problem of computing a sparse binary representation of an image. To be precise, given an image and an overcomplete, non-orthonormal basis, we aim to find a sparse binary vector indicating the minimal set of basis vectors that when added together best reconstruct the given input. We formulate this problem with an $L_2$ loss on the reconstruction error, and an $L_0$ (or, equivalently, an $L_1$) loss on the binary vector enforcing sparsity. This yields a so-called Quadratic Unconstrained Binary Optimization (QUBO) problem, whose solution is generally NP-hard to find. The contribution of this work is twofold. First, the method of unsupervised and unnormalized dictionary feature learning for a desired sparsity level to best match the data is presented. Second, the binary sparse coding problem is then solved on the Loihi 1 neuromorphic chip by the use of stochastic networks of neurons to traverse the non-convex energy landscape. The solutions are benchmarked against the classical heuristic simulated annealing. We demonstrate neuromorphic computing is suitable for sampling low energy solutions of binary sparse coding QUBO models, and although Loihi 1 is capable of sampling very sparse solutions of the QUBO models, there needs to be improvement in the implementation in order to be competitive with simulated annealing.

Elijah Pelofske, Vincent Russo

Quantum circuit unoptimization is an algorithm that transforms a quantum circuit into a different circuit that uses more gate operations while maintaining the same unitary transformation. We demonstrate that this method can implement digital zero-noise extrapolation (ZNE), a quantum error mitigation technique. By employing quantum circuit unoptimization as a form of circuit folding, noise can be systematically amplified. The key advantages of this approach are twofold. First, its ability to generate an exponentially increasing number of distinct circuit variants as the noise level is amplified, which allows noise averaging over many circuit variants with slightly different circuit structure. Averaging over these variants can mitigate the effect of biased error propagation due to the significantly altered circuit structure from quantum circuit unoptimization, or biased noise sources on a quantum processor. Second, quantum circuit unoptimization by design resists circuit simplification back to the original unmodified circuit, making it plausible to use ZNE in contexts where circuit compiler optimization is applied server-side. We evaluate the effectiveness of quantum circuit unoptimization as a noise-scaling method for ZNE in two test cases using depolarizing noise numerical simulations: random quantum volume circuits, where the observable is the heavy output probability, and QAOA circuits for the (unweighted) maximum cut problem on random 3-regular graphs, where the observable is the cut value. We show that using quantum circuit unoptimization to perform ZNE can approximately recover signal from noisy quantum simulations.

Elijah Pelofske, Vincent Urias, Lorie M. Liebrock

The task of accurate and efficient language translation is an extremely important information processing task. Machine learning enabled and automated translation that is accurate and fast is often a large topic of interest in the machine learning and data science communities. In this study, we examine using local Generative Pretrained Transformer (GPT) models to perform automated zero shot black-box, sentence wise, multi-natural-language translation into English text. We benchmark 16 different open-source GPT models, with no custom fine-tuning, from the Huggingface LLM repository for translating 50 different non-English languages into English using translated TED Talk transcripts as the reference dataset. These GPT model inference calls are performed strictly locally, on single A100 Nvidia GPUs. Benchmark metrics that are reported are language translation accuracy, using BLEU, GLEU, METEOR, and chrF text overlap measures, and wall-clock time for each sentence translation. The best overall performing GPT model for translating into English text for the BLEU metric is ReMM-v2-L2-13B with a mean score across all tested languages of $0.152$, for the GLEU metric is ReMM-v2-L2-13B with a mean score across all tested languages of $0.256$, for the chrF metric is Llama2-chat-AYT-13B with a mean score across all tested languages of $0.448$, and for the METEOR metric is ReMM-v2-L2-13B with a mean score across all tested languages of $0.438$.

Elijah Pelofske, Vincent Urias, Lorie M. Liebrock

Generative pre-trained transformers (GPT's) are a type of large language machine learning model that are unusually adept at producing novel, and coherent, natural language. In this study the ability of GPT models to generate novel and correct versions, and notably very insecure versions, of implementations of the cryptographic hash function SHA-1 is examined. The GPT models Llama-2-70b-chat-h, Mistral-7B-Instruct-v0.1, and zephyr-7b-alpha are used. The GPT models are prompted to re-write each function using a modified version of the localGPT framework and langchain to provide word embedding context of the full source code and header files to the model, resulting in over 150,000 function re-write GPT output text blocks, approximately 50,000 of which were able to be parsed as C code and subsequently compiled. The generated code is analyzed for being compilable, correctness of the algorithm, memory leaks, compiler optimization stability, and character distance to the reference implementation. Remarkably, several generated function variants have a high implementation security risk of being correct for some test vectors, but incorrect for other test vectors. Additionally, many function implementations were not correct to the reference algorithm of SHA-1, but produced hashes that have some of the basic characteristics of hash functions. Many of the function re-writes contained serious flaws such as memory leaks, integer overflows, out of bounds accesses, use of uninitialised values, and compiler optimization instability. Compiler optimization settings and SHA-256 hash checksums of the compiled binaries are used to cluster implementations that are equivalent but may not have identical syntax - using this clustering over 100,000 novel and correct versions of the SHA-1 codebase were generated where each component C function of the reference implementation is different from the original code.

Elijah Pelofske, Lorie M. Liebrock, Vincent Urias

In this research, we use user defined labels from three internet text sources (Reddit, Stackexchange, Arxiv) to train 21 different machine learning models for the topic classification task of detecting cybersecurity discussions in natural text. We analyze the false positive and false negative rates of each of the 21 model's in a cross validation experiment. Then we present a Cybersecurity Topic Classification (CTC) tool, which takes the majority vote of the 21 trained machine learning models as the decision mechanism for detecting cybersecurity related text. We also show that the majority vote mechanism of the CTC tool provides lower false negative and false positive rates on average than any of the 21 individual models. We show that the CTC tool is scalable to the hundreds of thousands of documents with a wall clock time on the order of hours.

Aaron Barbosa, Elijah Pelofske, Georg Hahn et al.

Quantum annealers, such as the device built by D-Wave Systems, Inc., offer a way to compute solutions of NP-hard problems that can be expressed in Ising or QUBO (quadratic unconstrained binary optimization) form. Although such solutions are typically of very high quality, problem instances are usually not solved to optimality due to imperfections of the current generations quantum annealers. In this contribution, we aim to understand some of the factors contributing to the hardness of a problem instance, and to use machine learning models to predict the accuracy of the D-Wave 2000Q annealer for solving specific problems. We focus on the Maximum Clique problem, a classic NP-hard problem with important applications in network analysis, bioinformatics, and computational chemistry. By training a machine learning classification model on basic problem characteristics such as the number of edges in the graph, or annealing parameters such as D-Wave's chain strength, we are able to rank certain features in the order of their contribution to the solution hardness, and present a simple decision tree which allows to predict whether a problem will be solvable to optimality with the D-Wave 2000Q. We extend these results by training a machine learning regression model that predicts the clique size found by D-Wave.

Aaron Barbosa, Elijah Pelofske, Georg Hahn et al.

Quantum annealers have been designed to propose near-optimal solutions to NP-hard optimization problems. However, the accuracy of current annealers such as the ones of D-Wave Systems, Inc., is limited by environmental noise and hardware biases. One way to deal with these imperfections and to improve the quality of the annealing results is to apply a variety of pre-processing techniques such as spin reversal (SR), anneal offsets (AO), or chain weights (CW). Maximizing the effectiveness of these techniques involves performing optimizations over a large number of parameters, which would be too costly if needed to be done for each new problem instance. In this work, we show that the aforementioned parameter optimization can be done for an entire class of problems, given each instance uses a previously chosen fixed embedding. Specifically, in the training phase, we fix an embedding E of a complete graph onto the hardware of the annealer, and then run an optimization algorithm to tune the following set of parameter values: the set of bits to be flipped for SR, the specific qubit offsets for AO, and the distribution of chain weights, optimized over a set of training graphs randomly chosen from that class, where the graphs are embedded onto the hardware using E. In the testing phase, we estimate how well the parameters computed during the training phase work on a random selection of other graphs from that class. We investigate graph instances of varying densities for the Maximum Clique, Maximum Cut, and Graph Partitioning problems. Our results indicate that, compared to their default behavior, substantial improvements of the annealing results can be achieved by using the optimized parameters for SR, AO, and CW.