MedTS: A BERT-based generation model to transform medical texts to SQL queries for electronic medical records (Preprint)

BACKGROUND Electronic medical records (EMRs) are usually stored in relational databases that require structured query language (SQL) queries to retrieve information of interest. Effectively completing such queries is usually a challenging task for medical experts due to the barriers in expertise. However, existing text-to-SQL generation studies have not been fully embraced in the medical domain. OBJECTIVE The objective of this study was to propose a neural generation model, which can jointly consider the characteristics of medical text and the SQL structure, to automatically transform medical texts to SQL queries for EMRs. METHODS In contrast to regarding the SQL query as an ordinary word sequence, the syntax tree, introduced as an intermediate representation, is more in line with the tree-structure nature of SQL and also can effectively reduce the search space during generation. We proposed a medical text-to-SQL model (MedTS), which employed a pre-trained BERT as the encoder and leveraged a grammar-based LSTM as the decoder to predict the tree-structured intermediate representation that can be easily transformed to the final SQL query. Experiments are conducted on the MIMICSQL dataset and five competitor methods are compared. RESULTS Experimental results demonstrated that MedTS achieved the accuracy of 0.770 and 0.888 on the test set in terms of logic form and execution respectively, which significantly outperformed the existing state-of-the-art methods. Further analyses proved that the performance on each component of the generated SQL was relatively balanced and has substantial improvements. CONCLUSIONS The proposed MedTS was effective and robust for improving the performance of medical text-to-SQL generation, indicating strong potentials to be applied in the real medical scenario.

Towards Version Controlling in RefactorErl

Static source code analyser tools are operating on an intermediate representation of the source code that is usually a tree or a graph. Those representations need to be updated according to the different versions of the source code. However, the developers might be interested in the changes or might need information about previous versions, therefore, keeping different versions of the source code analysed by the tools are required. RefactorErl is an open-source static analysis and transformation tool for Erlang that uses a graph representation to store and manipulate the source code. The aim of our research was to create an extension of the Semantic Program Graph of RefactorErl that is able to store different versions of the source code in a single graph. The new method resulted in 30% memory footprint decrease compared to the available workaround solutions.

Flynn’s Reconciliation

A large portion of the recent performance increase in the High Performance Computing (HPC) and Machine Learning (ML) domains is fueled by accelerator cards. Many popular ML frameworks support accelerators by organizing computations as a computational graph over a set of highly optimized, batched general-purpose kernels. While this approach simplifies the kernels’ implementation for each individual accelerator, the increasing heterogeneity among accelerator architectures for HPC complicates the creation of portable and extensible libraries of such kernels. Therefore, using a generalization of the CUDA community’s warp register cache programming idiom, we propose a new programming idiom (CoRe) and a virtual architecture model (PIRCH), abstracting over SIMD and SIMT paradigms. We define and automate the mapping process from a single source to PIRCH’s intermediate representation and develop backends that issue code for three different architectures: Intel AVX512, NVIDIA GPUs, and NEC SX-Aurora. Code generated by our source-to-source compiler for batched kernels, borG, competes favorably with vendor-tuned libraries and is up to 2× faster than hand-tuned kernels across architectures.

From IR Images to Point Clouds to Pose: Point Cloud-Based AR Glasses Pose Estimation

In this paper, we propose two novel AR glasses pose estimation algorithms from single infrared images by using 3D point clouds as an intermediate representation. Our first approach “PointsToRotation” is based on a Deep Neural Network alone, whereas our second approach “PointsToPose” is a hybrid model combining Deep Learning and a voting-based mechanism. Our methods utilize a point cloud estimator, which we trained on multi-view infrared images in a semi-supervised manner, generating point clouds based on one image only. We generate a point cloud dataset with our point cloud estimator using the HMDPose dataset, consisting of multi-view infrared images of various AR glasses with the corresponding 6-DoF poses. In comparison to another point cloud-based 6-DoF pose estimation named CloudPose, we achieve an error reduction of around 50%. Compared to a state-of-the-art image-based method, we reduce the pose estimation error by around 96%.

Automated Software Vulnerability Detection Based on Hybrid Neural Network

Vulnerabilities threaten the security of information systems. It is crucial to detect and patch vulnerabilities before attacks happen. However, existing vulnerability detection methods suffer from long-term dependency, out of vocabulary, bias towards global features or local features, and coarse detection granularity. This paper proposes an automatic vulnerability detection framework in source code based on a hybrid neural network. First, the inputs are transformed into an intermediate representation with explicit structure information using lower level virtual machine intermediate representation (LLVM IR) and backward program slicing. After the transformation, the size of samples and the size of vocabulary are significantly reduced. A hybrid neural network model is then applied to extract high-level features of vulnerability, which learns features both from convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The former is applied to learn local vulnerability features, such as buffer size. Furthermore, the latter is utilized to learn global features, such as data dependency. The extracted features are made up of concatenated outputs of CNN and RNN. Experiments are performed to validate our vulnerability detection method. The results show that our proposed method achieves excellent results with F1-scores of 98.6% and accuracy of 99.0% on the SARD dataset. It outperforms state-of-the-art methods.

Correct program parallelisations

A commonly used approach to develop deterministic parallel programs is to augment a sequential program with compiler directives that indicate which program blocks may potentially be executed in parallel. This paper develops a verification technique to reason about such compiler directives, in particular to show that they do not change the behaviour of the program. Moreover, the verification technique is tool-supported and can be combined with proving functional correctness of the program. To develop our verification technique, we propose a simple intermediate representation (syntax and semantics) that captures the main forms of deterministic parallel programs. This language distinguishes three kinds of basic blocks: parallel, vectorised and sequential blocks, which can be composed using three different composition operators: sequential, parallel and fusion composition. We show how a widely used subset of OpenMP can be encoded into this intermediate representation. Our verification technique builds on the notion of iteration contract to specify the behaviour of basic blocks; we show that if iteration contracts are manually specified for single blocks, then that is sufficient to automatically reason about data race freedom of the composed program. Moreover, we also show that it is sufficient to establish functional correctness on a linearised version of the original program to conclude functional correctness of the parallel program. Finally, we exemplify our approach on an example OpenMP program, and we discuss how tool support is provided.