Analysis of Failures and Risks in Deep Learning Model Converters: A Case Study in the ONNX Ecosystem
Software engineers develop, fine-tune, and deploy deep learning (DL) models. They use and re-use models in a variety of development frameworks and deploy them on a range of runtime environments. In this diverse ecosystem, engineers use DL model converters to move models from frameworks to runtime environments. However, errors in converters can compromise model quality and disrupt deployment. The failure frequency and failure modes of DL model converters are unknown. In this paper, we conduct the first failure analysis on DL model converters. Specifically, we characterize failures in model converters associated with ONNX (Open Neural Network eXchange). We analyze past failures in the ONNX converters in two major DL frameworks, PyTorch and TensorFlow. The symptoms, causes, and locations of failures (for N=200 issues), and trends over time are also reported. We also evaluate present-day failures by converting 8,797 models, both real-world and synthetically generated instances. The consistent result from both parts of the study is that DL model converters commonly fail by producing models that exhibit incorrect behavior: 33% of past failures and 8% of converted models fell into this category. Our results motivate future research on making DL software simpler to maintain, extend, and validate.
相關內容
Software engineering techniques are increasingly relying on deep learning approaches to support many software engineering tasks, from bug triaging to code generation. To assess the efficacy of such techniques researchers typically perform controlled experiments. Conducting these experiments, however, is particularly challenging given the complexity of the space of variables involved, from specialized and intricate architectures and algorithms to a large number of training hyper-parameters and choices of evolving datasets, all compounded by how rapidly the machine learning technology is advancing, and the inherent sources of randomness in the training process. In this work we conduct a mapping study, examining 194 experiments with techniques that rely on deep neural networks appearing in 55 papers published in premier software engineering venues to provide a characterization of the state-of-the-practice, pinpointing experiments common trends and pitfalls. Our study reveals that most of the experiments, including those that have received ACM artifact badges, have fundamental limitations that raise doubts about the reliability of their findings. More specifically, we find: weak analyses to determine that there is a true relationship between independent and dependent variables (87% of the experiments); limited control over the space of DNN relevant variables, which can render a relationship between dependent variables and treatments that may not be causal but rather correlational (100% of the experiments); and lack of specificity in terms of what are the DNN variables and their values utilized in the experiments (86% of the experiments) to define the treatments being applied, which makes it unclear whether the techniques designed are the ones being assessed, or how the sources of extraneous variation are controlled. We provide some practical recommendations to address these limitations.
Software design patterns present general code solutions to common software design problems. Modern software systems rely heavily on containers for organizing and orchestrating their constituent service components. Yet, despite the prevalence of ready-to-use Docker service images ready to participate in multi-container orchestration, developers do not have much guidance on how to develop their own multi-container Docker orchestrations. Thus in this work, we curate a dataset of successful projects that employ Docker Compose as an orchestration tool; then, we engage in qualitative and quantitative analysis of Docker Compose configurations. The collection of data and analysis enables the identification and naming of repeating patterns of deployment and orchestration employed by numerous successful open-source projects, much like software design patterns. These patterns highlight how software systems are orchestrated in the wild and can give examples to anybody wishing to develop their container orchestrations. These contributions also advance empirical research in software engineering patterns as evidence is provided about how Docker Compose is used.
Recent studies have explored the use of pre-trained embeddings for speech emotion recognition (SER), achieving comparable performance to conventional methods that rely on low-level knowledge-inspired acoustic features. These embeddings are often generated from models trained on large-scale speech datasets using self-supervised or weakly-supervised learning objectives. Despite the significant advancements made in SER through the use of pre-trained embeddings, there is a limited understanding of the trustworthiness of these methods, including privacy breaches, unfair performance, vulnerability to adversarial attacks, and computational cost, all of which may hinder the real-world deployment of these systems. In response, we introduce TrustSER, a general framework designed to evaluate the trustworthiness of SER systems using deep learning methods, with a focus on privacy, safety, fairness, and sustainability, offering unique insights into future research in the field of SER. Our code is publicly available under: //github.com/usc-sail/trust-ser.
WebAssembly (Wasm) is a novel low-level bytecode format that swiftly gained popularity for its efficiency, versatility and security, with near-native performance. Besides, trusted execution environments (TEEs) shield critical software assets against compromised infrastructures. However, TEEs do not guarantee the code to be trustworthy or that it was not tampered with. Instead, one relies on remote attestation to assess the code before execution. This paper describes WaTZ, which is (i) an efficient and secure runtime for trusted execution of Wasm code for Arm's TrustZone TEE, and (ii) a lightweight remote attestation system optimised for Wasm applications running in TrustZone, as it lacks built-in mechanisms for attestation. The remote attestation protocol is formally verified using a state-of-the-art analyser and model checker. Our extensive evaluation of Arm-based hardware uses synthetic and real-world benchmarks, illustrating typical tasks IoT devices achieve. WaTZ's execution speed is on par with Wasm runtimes in the normal world and reaches roughly half the speed of native execution, which is compensated by the additional security guarantees and the interoperability offered by Wasm. WaTZ is open-source and available on GitHub along with instructions to reproduce our experiments.
Scenarios are a crucial element for developing, testing, and verifying autonomous driving systems. However, open-source scenarios are often formulated using different terminologies. This limits their usage across different applications as many scenario representation formats are not directly compatible with each other. To address this problem, we present the first open-source converter from the OpenSCENARIO format to the CommonRoad format, which are two of the most popular scenario formats used in autonomous driving. Our converter employs a simulation tool to execute the dynamic elements defined by OpenSCENARIO. The converter is available at commonroad.in.tum.de and we demonstrate its usefulness by converting publicly available scenarios in the OpenSCENARIO format and evaluating them using CommonRoad tools.
With the extremely rapid advances in remote sensing (RS) technology, a great quantity of Earth observation (EO) data featuring considerable and complicated heterogeneity is readily available nowadays, which renders researchers an opportunity to tackle current geoscience applications in a fresh way. With the joint utilization of EO data, much research on multimodal RS data fusion has made tremendous progress in recent years, yet these developed traditional algorithms inevitably meet the performance bottleneck due to the lack of the ability to comprehensively analyse and interpret these strongly heterogeneous data. Hence, this non-negligible limitation further arouses an intense demand for an alternative tool with powerful processing competence. Deep learning (DL), as a cutting-edge technology, has witnessed remarkable breakthroughs in numerous computer vision tasks owing to its impressive ability in data representation and reconstruction. Naturally, it has been successfully applied to the field of multimodal RS data fusion, yielding great improvement compared with traditional methods. This survey aims to present a systematic overview in DL-based multimodal RS data fusion. More specifically, some essential knowledge about this topic is first given. Subsequently, a literature survey is conducted to analyse the trends of this field. Some prevalent sub-fields in the multimodal RS data fusion are then reviewed in terms of the to-be-fused data modalities, i.e., spatiospectral, spatiotemporal, light detection and ranging-optical, synthetic aperture radar-optical, and RS-Geospatial Big Data fusion. Furthermore, We collect and summarize some valuable resources for the sake of the development in multimodal RS data fusion. Finally, the remaining challenges and potential future directions are highlighted.
Trust has emerged as a key factor in people's interactions with AI-infused systems. Yet, little is known about what models of trust have been used and for what systems: robots, virtual characters, smart vehicles, decision aids, or others. Moreover, there is yet no known standard approach to measuring trust in AI. This scoping review maps out the state of affairs on trust in human-AI interaction (HAII) from the perspectives of models, measures, and methods. Findings suggest that trust is an important and multi-faceted topic of study within HAII contexts. However, most work is under-theorized and under-reported, generally not using established trust models and missing details about methods, especially Wizard of Oz. We offer several targets for systematic review work as well as a research agenda for combining the strengths and addressing the weaknesses of the current literature.
With the advent of 5G commercialization, the need for more reliable, faster, and intelligent telecommunication systems are envisaged for the next generation beyond 5G (B5G) radio access technologies. Artificial Intelligence (AI) and Machine Learning (ML) are not just immensely popular in the service layer applications but also have been proposed as essential enablers in many aspects of B5G networks, from IoT devices and edge computing to cloud-based infrastructures. However, most of the existing surveys in B5G security focus on the performance of AI/ML models and their accuracy, but they often overlook the accountability and trustworthiness of the models' decisions. Explainable AI (XAI) methods are promising techniques that would allow system developers to identify the internal workings of AI/ML black-box models. The goal of using XAI in the security domain of B5G is to allow the decision-making processes of the security of systems to be transparent and comprehensible to stakeholders making the systems accountable for automated actions. In every facet of the forthcoming B5G era, including B5G technologies such as RAN, zero-touch network management, E2E slicing, this survey emphasizes the role of XAI in them and the use cases that the general users would ultimately enjoy. Furthermore, we presented the lessons learned from recent efforts and future research directions on top of the currently conducted projects involving XAI.
Search in social networks such as Facebook poses different challenges than in classical web search: besides the query text, it is important to take into account the searcher's context to provide relevant results. Their social graph is an integral part of this context and is a unique aspect of Facebook search. While embedding-based retrieval (EBR) has been applied in eb search engines for years, Facebook search was still mainly based on a Boolean matching model. In this paper, we discuss the techniques for applying EBR to a Facebook Search system. We introduce the unified embedding framework developed to model semantic embeddings for personalized search, and the system to serve embedding-based retrieval in a typical search system based on an inverted index. We discuss various tricks and experiences on end-to-end optimization of the whole system, including ANN parameter tuning and full-stack optimization. Finally, we present our progress on two selected advanced topics about modeling. We evaluated EBR on verticals for Facebook Search with significant metrics gains observed in online A/B experiments. We believe this paper will provide useful insights and experiences to help people on developing embedding-based retrieval systems in search engines.
Deep neural networks (DNNs) are successful in many computer vision tasks. However, the most accurate DNNs require millions of parameters and operations, making them energy, computation and memory intensive. This impedes the deployment of large DNNs in low-power devices with limited compute resources. Recent research improves DNN models by reducing the memory requirement, energy consumption, and number of operations without significantly decreasing the accuracy. This paper surveys the progress of low-power deep learning and computer vision, specifically in regards to inference, and discusses the methods for compacting and accelerating DNN models. The techniques can be divided into four major categories: (1) parameter quantization and pruning, (2) compressed convolutional filters and matrix factorization, (3) network architecture search, and (4) knowledge distillation. We analyze the accuracy, advantages, disadvantages, and potential solutions to the problems with the techniques in each category. We also discuss new evaluation metrics as a guideline for future research.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191