Formal methods have enabled breakthroughs in many fields, such as in hardware verification, machine learning and biological systems. The key object of interest in systems biology, synthetic biology, and molecular programming is chemical reaction networks (CRNs) which formalizes coupled chemical reactions in a well-mixed solution. CRNs are pivotal for our understanding of biological regulatory and metabolic networks, as well as for programming engineered molecular behavior. Although it is clear that small CRNs are capable of complex dynamics and computational behavior, it remains difficult to explore the space of CRNs in search for desired functionality. We use Alloy, a tool for expressing structural constraints and behavior in software systems, to enumerate CRNs with declaratively specified properties. We show how this framework can enumerate CRNs with a variety of structural constraints including biologically motivated catalytic networks and metabolic networks, and seesaw networks motivated by DNA nanotechnology. We also use the framework to explore analog function computation in rate-independent CRNs. By computing the desired output value with stoichiometry rather than with reaction rates (in the sense that X → Y+Y computes multiplication by 2), such CRNs are completely robust to the choice of reaction rates or rate law. We find the smallest CRNs computing the max, minmax, abs and ReLU (rectified linear unit) functions in a natural subclass of rate-independent CRNs where rate-independence follows from structural network properties.

Embedding computation in molecular contexts incompatible with traditional electronics is expected to have wide ranging impact in synthetic biology, medicine, nanofabrication and other fields. A key remaining challenge lies in developing programming paradigms for molecular computation that are well-aligned with the underlying chemical hardware and do not attempt to shoehorn ill-fitting electronics paradigms. We discover a surprisingly tight connection between a popular class of neural networks (binary-weight ReLU aka BinaryConnect) and a class of coupled chemical reactions that are absolutely robust to reaction rates. The robustness of rate-independent chemical computation makes it a promising target for bioengineering implementation. We show how a BinaryConnect neural network trained in silico using well-founded deep learning optimization techniques, can be compiled to an equivalent chemical reaction network, providing a novel molecular programming paradigm. We illustrate such translation on the paradigmatic IRIS and MNIST datasets. Toward intended applications of chemical computation, we further use our method to generate a chemical reaction network that can discriminate between different virus types based on gene expression levels. Our work sets the stage for rich knowledge transfer between neural network and molecular programming communities.

As systems that deploy machine learning models become more and more pervasive, there is an urgent need to ensure their reliability and safety. While recent years have seen a lot of progress in techniques for verification and validation of machine learning models, reasoning about and explaining their behaviors remains challenging. This paper introduces a new approach for creating machine learning models where instead of the traditional supervised learning using data, the models are directly synthesized from specifications, and thus, are correct by construction. Our focus is binary classifiers with boolean features. Specifically, our approach translates relational specifications written in the well-known modeling language Alloy to neural networks that run on the widely used Tensorflow backend. Our key insight is that a slight enhancement of traditional boolean gates can provide a rich intermediate representation that readily translates to neural networks. To translate the enhanced gates to neural networks, we employ a state-of-the-art program synthesis framework that allows us to find minimal neural networks. The translation of Alloy specifications then follows the standard translation to boolean logic with the exception of utilizing the enhanced boolean gates, followed by a translation to neural networks. We embody our approach in a prototype tool and use it for experimental evaluation. The experimental results show that our approach allows synthesis of neural networks that are hard to create using traditional training methods.

This paper introduces the MCML approach for empirically studying the learnability of relational properties that can be expressed in the well-known software design language Alloy. A key novelty of MCML is quantification of the performance of and semantic differences among trained machine learning (ML) models, specifically decision trees, with respect to entire (bounded) input spaces, and not just for given training and test datasets (as is the common practice). MCML reduces the quantification problems to the classic complexity theory problem of model counting, and employs state-of-the-art model counters. The results show that relatively simple ML models can achieve surprisingly high performance (accuracy and F1-score) when evaluated in the common setting of using training and test datasets -- even when the training dataset is much smaller than the test dataset -- indicating the seeming simplicity of learning relational properties. However, MCML metrics based on model counting show that the performance can degrade substantially when tested against the entire (bounded) input space, indicating the high complexity of precisely learning these properties, and the usefulness of model counting in quantifying the true performance.

Modern software development processes recommend that changes be integrated into the main development line of a project multiple times a day. Before a new revision may be integrated, developers practice regression testing to ensure that the latest changes do not break any previously established functionality. The cost of regression testing is high, due to an increase in the number of revisions that are introduced per day, as well as the number of tests developers write per revision. Regression test selection (RTS) optimizes regression testing by skipping tests that are not affected by recent project changes. Existing dynamic RTS techniques support only projects written in a single programming language, which is unfortunate knowing that an open-source project is on average written in several programming languages.

Regression testing is used to check the correctness of evolving software. With the adoption of Agile development methodology, the number of tests and software revisions has dramatically increased, and hence has the cost of regression testing. Researchers proposed regression test selection (RTS) techniques that optimize regression testing by skipping tests that are not impacted by recent program changes. Ekstazi is one such state-of-the art technique; Ekstazi is implemented for the Java programming language and has been adopted by several companies and open-source projects.