PyTorchKR Papers

최신 논문

총 86편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2004.13873

Automated Physics-Derived Code Generation for Sensor Fusion and State Estimation

We present a new method for automatically generating the implementation of state-estimation algorithms from a machine-readable specification of the physics of a sensing system and physics of its signals and signal constraints. We implement the new state-estimator code generation method as a backend for a physics specification language and we apply the backend to generate complete C code implementations of state estimators for both linear systems (Kalman filters) and non-linear systems (extended Kalman filters). The state estimator code generation from physics specification is completely automated and requires no manual intervention. The generated filters can incorporate an Automatic Differentiation technique which combines function evaluation and differentiation in a single process. Using the description of physical system of a range of complexities, we generate extended Kalman filters, which we evaluate in terms of prediction accuracy using simulation traces. The results show that our automatically-generated sensor fusion and state estimation implementations provide state estimation within the same error bound as the human-written hand-optimized counterparts. We additionally quantify the code size and dynamic instruction count requirements of the generated state estimator implementations on the RISC-V architecture. The results show that our synthesized state estimation implementation employing Automatic Differentiation leads to an average improvement in the dynamic instruction count of the generated Kalman filter of 7%-16% compared to the standard differentiation technique. This is improvement comes at the limited cost of an average 4.5% increase in the code size of the generated filters.

arXiv:1811.11787

A Study of the Complexity and Accuracy of Direction of Arrival Estimation Methods Based on GCC-PHAT for a Pair of Close Microphones

This paper investigates the accuracy of various Generalized Cross-Correlation with Phase Transform (GCC-PHAT) methods for a close pair of microphones. We investigate interpolation-based methods and also propose another approach based on Singular Value Decomposition (SVD). All investigated methods are implemented in C code, and the execution time is measured to determine which approach is the most appealing for real-time applications on low-cost embedded hardware.

arXiv:1507.07039

High Performance Computing of Gene Regulatory Networks using a Message-Passing Model

Gene regulatory network reconstruction is a fundamental problem in computational biology. We recently developed an algorithm, called PANDA (Passing Attributes Between Networks for Data Assimilation), that integrates multiple sources of 'omics data and estimates regulatory network models. This approach was initially implemented in the C++ programming language and has since been applied to a number of biological systems. In our current research we are beginning to expand the algorithm to incorporate larger and most diverse data-sets, to reconstruct networks that contain increasing numbers of elements, and to build not only single network models, but sets of networks. In order to accomplish these "Big Data" applications, it has become critical that we increase the computational efficiency of the PANDA implementation. In this paper we show how to recast PANDA's similarity equations as matrix operations. This allows us to implement a highly readable version of the algorithm using the MATLAB/Octave programming language. We find that the resulting M-code much shorter (103 compared to 1128 lines) and more easily modifiable for potential future applications. The new implementation also runs significantly faster, with increasing efficiency as the network models increase in size. Tests comparing the C-code and M-code versions of PANDA demonstrate that this speed-up is on the order of 20-80 times faster for networks of similar dimensions to those we find in current biological applications.

arXiv:2008.11856공식 코드 · Papers with Code

Hybrid Deep Neural Networks to Infer State Models of Black-Box Systems

Inferring behavior model of a running software system is quite useful for several automated software engineering tasks, such as program comprehension, anomaly detection, and testing. Most existing dynamic model inference techniques are white-box, i.e., they require source code to be instrumented to get run-time traces. However, in many systems, instrumenting the entire source code is not possible (e.g., when using black-box third-party libraries) or might be very costly. Unfortunately, most black-box techniques that detect states over time are either univariate, or make assumptions on the data distribution, or have limited power for learning over a long period of past behavior. To overcome the above issues, in this paper, we propose a hybrid deep neural network that accepts as input a set of time series, one per input/output signal of the system, and applies a set of convolutional and recurrent layers to learn the non-linear correlations between signals and the patterns, over time. We have applied our approach on a real UAV auto-pilot solution from our industry partner with half a million lines of C code. We ran 888 random recent system-level test cases and inferred states, over time. Our comparison with several traditional time series change point detection techniques showed that our approach improves their performance by up to 102%, in terms of finding state change points, measured by F1 score. We also showed that our state classification algorithm provides on average 90.45% F1 score, which improves traditional classification algorithms by up to 17%.

arXiv:2008.11830

Designing Neural Networks for Real-Time Systems

Artificial Neural Networks (ANNs) are increasingly being used within safety-critical Cyber-Physical Systems (CPSs). They are often co-located with traditional embedded software, and may perform advisory or control-based roles. It is important to validate both the timing and functional correctness of these systems. However, most approaches in the literature consider guaranteeing only the functionality of ANN based controllers. This issue stems largely from the implementation strategies used within common neural network frameworks -- their underlying source code is often simply unsuitable for formal techniques such as static timing analysis. As a result, developers of safety-critical CPS must rely on informal techniques such as measurement based approaches to prove correctness, techniques that provide weak guarantees at best. In this work we address this challenge. We propose a design pipeline whereby neural networks trained using the popular deep learning framework Keras are compiled to functionally equivalent C code. This C code is restricted to simple constructs that may be analysed by existing static timing analysis tools. As a result, if compiled to a suitable time-predictable platform all execution bounds may be statically derived. To demonstrate the benefits of our approach we execute an ANN trained to drive an autonomous vehicle around a race track. We compile the ANN to the Patmos time-predictable controller, and show that we can derive worst case execution timings.

arXiv:2008.07127공식 코드 · Papers with Code

DORY: Automatic End-to-End Deployment of Real-World DNNs on Low-Cost IoT MCUs

The deployment of Deep Neural Networks (DNNs) on end-nodes at the extreme edge of the Internet-of-Things is a critical enabler to support pervasive Deep Learning-enhanced applications. Low-Cost MCU-based end-nodes have limited on-chip memory and often replace caches with scratchpads, to reduce area overheads and increase energy efficiency -- requiring explicit DMA-based memory transfers between different levels of the memory hierarchy. Mapping modern DNNs on these systems requires aggressive topology-dependent tiling and double-buffering. In this work, we propose DORY (Deployment Oriented to memoRY) - an automatic tool to deploy DNNs on low cost MCUs with typically less than 1MB of on-chip SRAM memory. DORY abstracts tiling as a Constraint Programming (CP) problem: it maximizes L1 memory utilization under the topological constraints imposed by each DNN layer. Then, it generates ANSI C code to orchestrate off- and on-chip transfers and computation phases. Furthermore, to maximize speed, DORY augments the CP formulation with heuristics promoting performance-effective tile sizes. As a case study for DORY, we target GreenWaves Technologies GAP8, one of the most advanced parallel ultra-low power MCU-class devices on the market. On this device, DORY achieves up to 2.5x better MAC/cycle than the GreenWaves proprietary software solution and 18.1x better than the state-of-the-art result on an STM32-F746 MCU on single layers. Using our tool, GAP-8 can perform end-to-end inference of a 1.0-MobileNet-128 network consuming just 63 pJ/MAC on average @ 4.3 fps - 15.4x better than an STM32-F746. We release all our developments - the DORY framework, the optimized backend kernels, and the related heuristics - as open-source software.

arXiv:2007.06708공식 코드 · Papers with Code

CPL-SLAM: Efficient and Certifiably Correct Planar Graph-Based SLAM Using the Complex Number Representation

In this paper, we consider the problem of planar graph-based simultaneous localization and mapping (SLAM) that involves both poses of the autonomous agent and positions of observed landmarks. We present CPL-SLAM, an efficient and certifiably correct algorithm to solve planar graph-based SLAM using the complex number representation. We formulate and simplify planar graph-based SLAM as the maximum likelihood estimation (MLE) on the product of unit complex numbers, and relax this nonconvex quadratic complex optimization problem to convex complex semidefinite programming (SDP). Furthermore, we simplify the corresponding complex semidefinite programming to Riemannian staircase optimization (RSO) on the complex oblique manifold that can be solved with the Riemannian trust region (RTR) method. In addition, we prove that the SDP relaxation and RSO simplification are tight as long as the noise magnitude is below a certain threshold. The efficacy of this work is validated through applications of CPL-SLAM and comparisons with existing state-of-the-art methods on planar graph-based SLAM, which indicates that our proposed algorithm is capable of solving planar graph-based SLAM certifiably, and is more efficient in numerical computation and more robust to measurement noise than existing state-of-the-art methods. The C++ code for CPL-SLAM is available at https://github.com/MurpheyLab/CPL-SLAM.

arXiv:2006.09973

Breaking Type Safety in Go: An Empirical Study on the Usage of the unsafe Package

A decade after its first release, the Go programming language has become a major programming language in the development landscape. While praised for its clean syntax and C-like performance, Go also contains a strong static type-system that prevents arbitrary type casting and arbitrary memory access, making the language type-safe by design. However, to give developers the possibility of implementing low-level code, Go ships with a special package called unsafe that offers developers a way around the type-safety of Go programs. The package gives greater flexibility to developers but comes at a higher risk of runtime errors, chances of non-portability, and the loss of compatibility guarantees for future versions of Go. In this paper, we present the first large-scale study on the usage of the unsafe package in 2,438 popular Go projects. Our investigation shows that unsafe is used in 24% of Go projects, motivated primarily by communicating with operating systems and C code, but is also commonly used as a source of performance optimization. Developers are willing to use unsafe to break language specifications (e.g., string immutability) for better performance and 6% of analyzed projects that use unsafe perform risky pointer conversions that can lead to program crashes and unexpected behavior. Furthermore, we report a series of real issues faced by projects that use unsafe, from crashing errors and non-deterministic behavior to having their deployment restricted from certain popular environments. Our findings can be used to understand how and why developers break type-safety in Go, and help motivate further tools and language development that could make the usage of unsafe in Go even safer.

arXiv:2006.03250

FP-Stereo: Hardware-Efficient Stereo Vision for Embedded Applications

Fast and accurate depth estimation, or stereo matching, is essential in embedded stereo vision systems, requiring substantial design effort to achieve an appropriate balance among accuracy, speed and hardware cost. To reduce the design effort and achieve the right balance, we propose FP-Stereo for building high-performance stereo matching pipelines on FPGAs automatically. FP-Stereo consists of an open-source hardware-efficient library, allowing designers to obtain the desired implementation instantly. Diverse methods are supported in our library for each stage of the stereo matching pipeline and a series of techniques are developed to exploit the parallelism and reduce the resource overhead. To improve the usability, FP-Stereo can generate synthesizable C code of the FPGA accelerator with our optimized HLS templates automatically. To guide users for the right design choice meeting specific application requirements, detailed comparisons are performed on various configurations of our library to investigate the accuracy/speed/cost trade-off. Experimental results also show that FP-Stereo outperforms the state-of-the-art FPGA design from all aspects, including 6.08% lower error, 2x faster speed, 30% less resource usage and 40% less energy consumption. Compared to GPU designs, FP-Stereo achieves the same accuracy at a competitive speed while consuming much less energy.

논문 ID: publication/3413100

Simulated Annealing Algorithm for the Multiple Choice Multidimensional Knapsack Problem

The multiple choice multidimensional knapsack problem (MCMK) is a harder version of the 0/1 knapsack problem, and is ever more complex than the 0/1 multidimensional knapsack problem. In MCMK, there are several groups of items. The objective is to maximize the value (profit) by choosing exactly 1 item from each group such that all the constraints are satisfied. It is difficult and NP-hard even to find a solution that does not violate all constraints. In this work, we present a simulated annealing based algorithm with open source C++ code to find good solutions to the multidimensional multiple choice knapsack problem. In all of the benchmark instances we used, the algorithm is able to find optimum (or close) solutions, thereby proving that the algorithm is suitable for solving larger instances for which optimal solutions are unknown.

arXiv:2002.03869

A Framework for Semi-Automatic Precision and Accuracy Analysis for Fast and Rigorous Deep Learning

Deep Neural Networks (DNN) represent a performance-hungry application. Floating-Point (FP) and custom floating-point-like arithmetic satisfies this hunger. While there is need for speed, inference in DNNs does not seem to have any need for precision. Many papers experimentally observe that DNNs can successfully run at almost ridiculously low precision. The aim of this paper is two-fold: first, to shed some theoretical light upon why a DNN's FP accuracy stays high for low FP precision. We observe that the loss of relative accuracy in the convolutional steps is recovered by the activation layers, which are extremely well-conditioned. We give an interpretation for the link between precision and accuracy in DNNs. Second, the paper presents a software framework for semi-automatic FP error analysis for the inference phase of deep-learning. Compatible with common Tensorflow/Keras models, it leverages the frugally-deep Python/C++ library to transform a neural network into C++ code in order to analyze the network's need for precision. This rigorous analysis is based on Interval and Affine arithmetics to compute absolute and relative error bounds for a DNN. We demonstrate our tool with several examples.

arXiv:2001.05572

A C Code Generator for Fast Inference and Simple Deployment of Convolutional Neural Networks on Resource Constrained Systems

Inference of Convolutional Neural Networks in time critical applications usually requires a GPU. In robotics or embedded devices these are often not available due to energy, space and cost constraints. Furthermore, installation of a deep learning framework or even a native compiler on the target platform is not possible. This paper presents a neural network code generator (NNCG) that generates from a trained CNN a plain ANSI C code file that encapsulates the inference in single a function. It can easily be included in existing projects and due to lack of dependencies, cross compilation is usually possible. Additionally, the code generation is optimized based on the known trained CNN and target platform following four design principles. The system is evaluated utilizing small CNN designed for this application. Compared to TensorFlow XLA and Glow speed-ups of up to 11.81 can be shown and even GPUs are outperformed regarding latency.

arXiv:2001.03541

Multi-layer Optimizations for End-to-End Data Analytics

We consider the problem of training machine learning models over multi-relational data. The mainstream approach is to first construct the training dataset using a feature extraction query over input database and then use a statistical software package of choice to train the model. In this paper we introduce Iterative Functional Aggregate Queries (IFAQ), a framework that realizes an alternative approach. IFAQ treats the feature extraction query and the learning task as one program given in the IFAQ's domain-specific language, which captures a subset of Python commonly used in Jupyter notebooks for rapid prototyping of machine learning applications. The program is subject to several layers of IFAQ optimizations, such as algebraic transformations, loop transformations, schema specialization, data layout optimizations, and finally compilation into efficient low-level C++ code specialized for the given workload and data. We show that a Scala implementation of IFAQ can outperform mlpack, Scikit, and TensorFlow by several orders of magnitude for linear regression and regression tree models over several relational datasets.

arXiv:1910.00577공식 코드 · Papers with Code

Structural Language Models of Code

We address the problem of any-code completion - generating a missing piece of source code in a given program without any restriction on the vocabulary or structure. We introduce a new approach to any-code completion that leverages the strict syntax of programming languages to model a code snippet as a tree - structural language modeling (SLM). SLM estimates the probability of the program's abstract syntax tree (AST) by decomposing it into a product of conditional probabilities over its nodes. We present a neural model that computes these conditional probabilities by considering all AST paths leading to a target node. Unlike previous techniques that have severely restricted the kinds of expressions that can be generated in this task, our approach can generate arbitrary code in any programming language. Our model significantly outperforms both seq2seq and a variety of structured approaches in generating Java and C# code. Our code, data, and trained models are available at http://github.com/tech-srl/slm-code-generation/ . An online demo is available at http://AnyCodeGen.org .

arXiv:1908.08022공식 코드 · Papers with Code

Genetic Algorithm for the 0/1 Multidimensional Knapsack Problem

The 0/1 multidimensional knapsack problem is the 0/1 knapsack problem with m constraints which makes it difficult to solve using traditional methods like dynamic programming or branch and bound algorithms. We present a genetic algorithm for the multidimensional knapsack problem with Java and C++ code that is able to solve publicly available instances in a very short computational duration. Our algorithm uses iteratively computed Lagrangian multipliers as constraint weights to augment the greedy algorithm for the multidimensional knapsack problem and uses that information in a greedy crossover in a genetic algorithm. The algorithm uses several other hyperparameters which can be set in the code to control convergence. Our algorithm improves upon the algorithm by Chu and Beasley in that it converges to optimum or near optimum solutions much faster.

arXiv:1605.01421

SimInf: An R package for Data-driven Stochastic Disease Spread Simulations

We present the R package SimInf which provides an efficient and very flexible framework to conduct data-driven epidemiological modeling in realistic large scale disease spread simulations. The framework integrates infection dynamics in subpopulations as continuous-time Markov chains using the Gillespie stochastic simulation algorithm and incorporates available data such as births, deaths and movements as scheduled events at predefined time-points. Using C code for the numerical solvers and OpenMP to divide work over multiple processors ensures high performance when simulating a sample outcome. One of our design goal was to make SimInf extendable and enable usage of the numerical solvers from other R extension packages in order to facilitate complex epidemiological research. In this paper, we provide a technical description of the framework and demonstrate its use on some basic examples. We also discuss how to specify and extend the framework with user-defined models.

arXiv:1905.08325

Towards Neural Decompilation

We address the problem of automatic decompilation, converting a program in low-level representation back to a higher-level human-readable programming language. The problem of decompilation is extremely important for security researchers. Finding vulnerabilities and understanding how malware operates is much easier when done over source code. The importance of decompilation has motivated the construction of hand-crafted rule-based decompilers. Such decompilers have been designed by experts to detect specific control-flow structures and idioms in low-level code and lift them to source level. The cost of supporting additional languages or new language features in these models is very high. We present a novel approach to decompilation based on neural machine translation. The main idea is to automatically learn a decompiler from a given compiler. Given a compiler from a source language S to a target language T , our approach automatically trains a decompiler that can translate (decompile) T back to S . We used our framework to decompile both LLVM IR and x86 assembly to C code with high success rates. Using our LLVM and x86 instantiations, we were able to successfully decompile over 97% and 88% of our benchmarks respectively.

arXiv:1904.04421

FPGA/DNN Co-Design: An Efficient Design Methodology for IoT Intelligence on the Edge

While embedded FPGAs are attractive platforms for DNN acceleration on edge-devices due to their low latency and high energy efficiency, the scarcity of resources of edge-scale FPGA devices also makes it challenging for DNN deployment. In this paper, we propose a simultaneous FPGA/DNN co-design methodology with both bottom-up and top-down approaches: a bottom-up hardware-oriented DNN model search for high accuracy, and a top-down FPGA accelerator design considering DNN-specific characteristics. We also build an automatic co-design flow, including an Auto-DNN engine to perform hardware-oriented DNN model search, as well as an Auto-HLS engine to generate synthesizable C code of the FPGA accelerator for explored DNNs. We demonstrate our co-design approach on an object detection task using PYNQ-Z1 FPGA. Results show that our proposed DNN model and accelerator outperform the state-of-the-art FPGA designs in all aspects including Intersection-over-Union (IoU) (6.2% higher), frames per second (FPS) (2.48X higher), power consumption (40% lower), and energy efficiency (2.5X higher). Compared to GPU-based solutions, our designs deliver similar accuracy but consume far less energy.

arXiv:1206.6466

Utilizing Static Analysis and Code Generation to Accelerate Neural Networks

As datasets continue to grow, neural network (NN) applications are becoming increasingly limited by both the amount of available computational power and the ease of developing high-performance applications. Researchers often must have expert systems knowledge to make their algorithms run efficiently. Although available computing power increases rapidly each year, algorithm efficiency is not able to keep pace due to the use of general purpose compilers, which are not able to fully optimize specialized application domains. Within the domain of NNs, we have the added knowledge that network architecture remains constant during training, meaning the architecture's data structure can be statically optimized by a compiler. In this paper, we present SONNC, a compiler for NNs that utilizes static analysis to generate optimized parallel code. We show that SONNC's use of static optimizations make it able to outperform hand-optimized C++ code by up to 7.8X, and MATLAB code by up to 24X. Additionally, we show that use of SONNC significantly reduces code complexity when using structurally sparse networks.