Logs are a common way to record detailed run-time information in software. As modern software systems evolve in scale and complexity, logs have become indispensable to understanding the internal states of the system. At the same time however, manually inspecting logs has become impractical. In recent times, there has been more emphasis on statistical and machine learning (ML) based methods for analyzing logs. While the results have shown promise, most of the literature focuses on algorithms and state-of-the-art (SOTA), while largely ignoring the practical aspects. In this paper we demonstrate our end-to-end log classification pipeline, Linnaeus. Besides showing the more traditional ML flow, we also demonstrate our solutions for adaptability and re-use, integration towards large scale software development processes, and how we cope with lack of labelled data. We hope Linnaeus can serve as a blueprint for, and inspire the integration of, various ML based solutions in other large scale industrial settings.
Fall prevalence is high among elderly people, which is challenging due to the severe consequences of falling. This is why rapid assistance is a critical task. Ambient assisted living (AAL) uses recent technologies such as 5G networks and the internet of medical things (IoMT) to address this research area. Edge computing can reduce the cost of cloud communication, including high latency and bandwidth use, by moving conventional healthcare services and applications closer to end-users. Artificial intelligence (AI) techniques such as deep learning (DL) have been used recently for automatic fall detection, as well as supporting healthcare services. However, DL requires a vast amount of data and substantial processing power to improve its performance for the IoMT linked to the traditional edge computing environment. This research proposes an effective fall detection framework based on DL algorithms and mobile edge computing (MEC) within 5G wireless networks, the aim being to empower IoMT-based healthcare applications. We also propose the use of a deep gated recurrent unit (DGRU) neural network to improve the accuracy of existing DL-based fall detection methods. DGRU has the advantage of dealing with time-series IoMT data, and it can reduce the number of parameters and avoid the vanishing gradient problem. The experimental results on two public datasets show that the DGRU model of the proposed framework achieves higher accuracy rates compared to the current related works on the same datasets.
For the past six years, researchers in genetic programming and other program synthesis disciplines have used the General Program Synthesis Benchmark Suite to benchmark many aspects of automatic program synthesis systems. These problems have been used to make notable progress toward the goal of general program synthesis: automatically creating the types of software that human programmers code. Many of the systems that have attempted the problems in the original benchmark suite have used it to demonstrate performance improvements granted through new techniques. Over time, the suite has gradually become outdated, hindering the accurate measurement of further improvements. The field needs a new set of more difficult benchmark problems to move beyond what was previously possible. In this paper, we describe the 25 new general program synthesis benchmark problems that make up PSB2, a new benchmark suite. These problems are curated from a variety of sources, including programming katas and college courses. We selected these problems to be more difficult than those in the original suite, and give results using PushGP showing this increase in difficulty. These new problems give plenty of room for improvement, pointing the way for the next six or more years of general program synthesis research.
The factual knowledge acquired during pretraining and stored in the parameters of Language Models (LM) can be useful in downstream tasks (e.g., question answering or textual inference). However, some facts can be incorrectly induced or become obsolete over time. We present KnowledgeEditor, a method that can be used to edit this knowledge and, thus, fix 'bugs' or unexpected predictions without the need for expensive re-training or fine-tuning. Besides being computationally efficient, KnowledgeEditor does not require any modifications in LM pre-training (e.g., the use of meta-learning). In our approach, we train a hyper-network with constrained optimization to modify a fact without affecting the rest of the knowledge; the trained hyper-network is then used to predict the weight update at test time. We show KnowledgeEditor's efficacy with two popular architectures and knowledge-intensive tasks: i) a BERT model fine-tuned for fact-checking, and ii) a sequence-to-sequence BART model for question answering. With our method, changing a prediction on the specific wording of a query tends to result in a consistent change in predictions also for its paraphrases. We show that this can be further encouraged by exploiting (e.g., automatically-generated) paraphrases during training. Interestingly, our hyper-network can be regarded as a 'probe' revealing which components of a model need to be changed to manipulate factual knowledge; our analysis shows that the updates tend to be concentrated on a small subset of components. Code at https://github.com/nicola-decao/KnowledgeEditor
Semantic segmentation approaches are typically trained on large-scale data with a closed finite set of known classes without considering unknown objects. In certain safety-critical robotics applications, especially autonomous driving, it is important to segment all objects, including those unknown at training time. We formalize the task of video class agnostic segmentation from monocular video sequences in autonomous driving to account for unknown objects. Video class agnostic segmentation can be formulated as an open-set or a motion segmentation problem. We discuss both formulations and provide datasets and benchmark different baseline approaches for both tracks. In the motion-segmentation track we benchmark real-time joint panoptic and motion instance segmentation, and evaluate the effect of ego-flow suppression. In the open-set segmentation track we evaluate baseline methods that combine appearance, and geometry to learn prototypes per semantic class. We then compare it to a model that uses an auxiliary contrastive loss to improve the discrimination between known and unknown objects. All datasets and models are publicly released at https://msiam.github.io/vca/.
Avian botulism caused by a bacterium, Clostridium botulinum, causes a paralytic disease in birds often leading to high fatality, and is usually diagnosed using molecular techniques. Diagnostic techniques for Avian botulism include: Mouse Bioassay, ELISA, PCR, all of which are time-consuming, laborious and require invasive sample collection from affected sites. In this study, we build a first-ever multi-spectral, remote-sensing imagery based global Bird-Area Water-bodies Dataset (BAWD) (i.e. fused satellite images of water-body sites important for avian fauna) backed by on-ground reporting evidence of outbreaks. In the current version, BAWD covers a total ground area of 904 sq.km from two open source satellite projects (Sentinel and Landsat). BAWD consists of 17 topographically diverse global sites spanning across 4 continents, with locations monitored over a time-span of 3 years (2016-2020). Using BAWD and state-of-the-art deep-learning techniques we propose a first-ever Artificial Intelligence based (AI) model to predict potential outbreak of Avian botulism called AVI-BoT (Aerosol, Visible, Infra-red (NIR/SWIR) and Bands of Thermal). AVI-BoT uses fused multi-spectral satellite images of water-bodies (10-bands) as input to generate a spatial prediction map depicting probability of potential Avian botulism outbreaks. We also train and investigate a simpler (5-band) Causative-Factor model (based on prominent physiological factors reported in literature as conducive for outbreak) to predict Avian botulism. Using AVI-BoT, we achieve a training accuracy of 0.94 and validation accuracy of 0.96 on BAWD, far superior in comparison to our Causative factors model. The proposed technique presents a scale-able, low-cost, non-invasive methodology for continuous monitoring of bird-habitats against botulism outbreaks with the potential of saving valuable fauna lives.
Elasticity image, visualizing the quantitative map of tissue stiffness, can be reconstructed by solving an inverse problem. Classical methods for magnetic resonance elastography (MRE) try to solve a regularized optimization problem comprising a deterministic physical model and a prior constraint as data-fidelity term and regularization term, respectively. For improving the elasticity reconstructions, appropriate prior about the underlying elasticity distribution is required which is not unique. This article proposes an infused approach for MRE reconstruction by integrating the statistical representation of the physical laws of harmonic motions and learning-based prior. For data-fidelity term, we use a statistical linear-algebraic model of equilibrium equations and for the regularizer, data-driven regularization by denoising (RED) is utilized. In the proposed optimization paradigm, the regularizer gradient is simply replaced by the residual of learned denoiser leading to time-efficient computation and convex explicit objective function. Simulation results of elasticity reconstruction verify the effectiveness of the proposed approach.
This paper studies the optimal state estimation problem for interconnected systems. Each subsystem can obtain its own measurement in real time, while, the measurements transmitted between the subsystems suffer from random delay. The optimal estimator is analytically designed for minimizing the conditional error covariance. Due to the random delay, the error covariance of the estimation is random. The boundedness of the expected error covariance (EEC) is analyzed. In particular, a new condition that is easy to verify is established for the boundedness of EEC. Further, the properties about EEC with respect to the delay probability is studied. We found that there exists a critical probability such that the EEC is bounded if the delay probability is below the critical probability. Also, a lower and upper bound of the critical probability is effectively computed. Finally, the proposed results are applied to a power system, and the effectiveness of the designed methods is illustrated by simulations.
Prior work demonstrated the ability of machine learning to automatically recognize surgical workflow steps from videos. However, these studies focused on only a single type of procedure. In this work, we analyze, for the first time, surgical step recognition on four different laparoscopic surgeries: Cholecystectomy, Right Hemicolectomy, Sleeve Gastrectomy, and Appendectomy. Inspired by the traditional apprenticeship model, in which surgical training is based on the Halstedian method, we paraphrase the "see one, do one, teach one" approach for the surgical intelligence domain as "train one, classify one, teach one". In machine learning, this approach is often referred to as transfer learning. To analyze the impact of transfer learning across different laparoscopic procedures, we explore various time-series architectures and examine their performance on each target domain. We introduce a new architecture, the Time-Series Adaptation Network (TSAN), an architecture optimized for transfer learning of surgical step recognition, and we show how TSAN can be pre-trained using self-supervised learning on a Sequence Sorting task. Such pre-training enables TSAN to learn workflow steps of a new laparoscopic procedure type from only a small number of labeled samples from the target procedure. Our proposed architecture leads to better performance compared to other possible architectures, reaching over 90% accuracy when transferring from laparoscopic Cholecystectomy to the other three procedure types.
Batch Normalization (BN) is a popular technique for training Deep Neural Networks (DNNs). BN uses scaling and shifting to normalize activations of mini-batches to accelerate convergence and improve generalization. The recently proposed Iterative Normalization (IterNorm) method improves these properties by whitening the activations iteratively using Newton's method. However, since Newton's method initializes the whitening matrix independently at each training step, no information is shared between consecutive steps. In this work, instead of exact computation of whitening matrix at each time step, we estimate it gradually during training in an online fashion, using our proposed Stochastic Whitening Batch Normalization (SWBN) algorithm. We show that while SWBN improves the convergence rate and generalization of DNNs, its computational overhead is less than that of IterNorm. Due to the high efficiency of the proposed method, it can be easily employed in most DNN architectures with a large number of layers. We provide comprehensive experiments and comparisons between BN, IterNorm, and SWBN layers to demonstrate the effectiveness of the proposed technique in conventional (many-shot) image classification and few-shot classification tasks.