Search

We propose a hybrid framework for consistently producing high-quality object tracks by combining an automated object tracker with little human input. The key idea is to tailor a module for each dataset to intelligently decide when an object tracker is failing and so humans should be brought in to re-localize an object for continued tracking. Our approach leverages self-supervised learning on unlabeled videos to learn a tailored representation for a target object that is then used to actively monitor its tracked region and decide when the tracker fails. Since labeled data is not needed, our approach can be applied to novel object categories. Experiments on three datasets demonstrate our method outperforms existing approaches, especially for small, fast moving, or occluded objects.

Visual question answering is a task of predicting the answer to a question about an image. Given that different people can provide different answers to a visual question, we aim to better understand why with answer groundings. We introduce the first dataset that visually grounds each unique answer to each visual question, which we call VQAAnswerTherapy. We then propose two novel problems of predicting whether a visual question has a single answer grounding and localizing all answer groundings. We benchmark modern algorithms for these novel problems to show where they succeed and struggle. The dataset and evaluation server can be found publicly at https://vizwiz.org/tasks-and-datasets/vqa-answer-therapy/., Comment: IEEE/CVF International Conference on Computer Vision

The emergence of large language models has led to the development of powerful tools such as ChatGPT that can produce text indistinguishable from human-generated work. With the increasing accessibility of such technology, students across the globe may utilize it to help with their school work -- a possibility that has sparked discussions on the integrity of student evaluations in the age of artificial intelligence (AI). To date, it is unclear how such tools perform compared to students on university-level courses. Further, students' perspectives regarding the use of such tools, and educators' perspectives on treating their use as plagiarism, remain unknown. Here, we compare the performance of ChatGPT against students on 32 university-level courses. We also assess the degree to which its use can be detected by two classifiers designed specifically for this purpose. Additionally, we conduct a survey across five countries, as well as a more in-depth survey at the authors' institution, to discern students' and educators' perceptions of ChatGPT's use. We find that ChatGPT's performance is comparable, if not superior, to that of students in many courses. Moreover, current AI-text classifiers cannot reliably detect ChatGPT's use in school work, due to their propensity to classify human-written answers as AI-generated, as well as the ease with which AI-generated text can be edited to evade detection. Finally, we find an emerging consensus among students to use the tool, and among educators to treat this as plagiarism. Our findings offer insights that could guide policy discussions addressing the integration of AI into educational frameworks., Comment: 17 pages, 4 figures

This research study explores the use of near-miss clone detection to support the characterization of domain models of smart contracts for each of the popular domains in which smart contracts are being rapidly adopted. In this paper, we leverage the code clone detection techniques to detect similarities in functions of the smart contracts deployed onto the Ethereum blockchain network. We analyze the clusters of code clones and the semantics of the code fragments in the clusters in an attempt to categorize them and discover the structural models of the patterns in code clones.

Ethereum Smart Contracts based on Blockchain Technology (BT) enables monetary transactions among peers on a blockchain network independent of a central authorizing agency. Ethereum Smart Contracts are programs that are deployed as decentralized applications, having the building blocks of the blockchain consensus protocol. This enables consumers to make agreements in a transparent and conflict-free environment. However, there exist some security vulnerabilities within these smart contracts that are a potential threat to the applications and their consumers and have shown in the past to cause huge financial losses. This paper presents a framework and empirical analysis that use code clone detection techniques for identifying vulnerabilities and their variations in smart contracts. Our empirical analysis is conducted using the Nicad code clone detection tool on a dataset of approximately 50k Ethereum smart contracts. We evaluated VOLCANO on two datasets, one with confirmed vulnerabilities and another with approximately 50k random smart contracts collected from the Etherscan. Our approach shows an improvement in the detection of vulnerabilities in terms of coverage and efficiency when compared to two of the publicly available static analyzers to detect vulnerabilities in smart contracts. To the best of our knowledge, this is the first study that uses a clone detection technique to identify vulnerabilities and their evolution in Ethereum smart contracts.

Visual question answering is the task of answering questions about images. We introduce the VizWiz-VQA-Grounding dataset, the first dataset that visually grounds answers to visual questions asked by people with visual impairments. We analyze our dataset and compare it with five VQA-Grounding datasets to demonstrate what makes it similar and different. We then evaluate the SOTA VQA and VQA-Grounding models and demonstrate that current SOTA algorithms often fail to identify the correct visual evidence where the answer is located. These models regularly struggle when the visual evidence occupies a small fraction of the image, for images that are higher quality, as well as for visual questions that require skills in text recognition. The dataset, evaluation server, and leaderboard all can be found at the following link: https://vizwiz.org/tasks-and-datasets/answer-grounding-for-vqa/., Comment: Computer Vision and Pattern Recognition

In recent years, low-resource Machine Reading Comprehension (MRC) has made significant progress, with models getting remarkable performance on various language datasets. However, none of these models have been customized for the Urdu language. This work explores the semi-automated creation of the Urdu Question Answering Dataset (UQuAD1.0) by combining machine-translated SQuAD with human-generated samples derived from Wikipedia articles and Urdu RC worksheets from Cambridge O-level books. UQuAD1.0 is a large-scale Urdu dataset intended for extractive machine reading comprehension tasks consisting of 49k question Answers pairs in question, passage, and answer format. In UQuAD1.0, 45000 pairs of QA were generated by machine translation of the original SQuAD1.0 and approximately 4000 pairs via crowdsourcing. In this study, we used two types of MRC models: rule-based baseline and advanced Transformer-based models. However, we have discovered that the latter outperforms the others; thus, we have decided to concentrate solely on Transformer-based architectures. Using XLMRoBERTa and multi-lingual BERT, we acquire an F1 score of 0.66 and 0.63, respectively., Comment: 22 pages, 6 figures

Ethereum Smart Contracts based on Blockchain Technology (BT)enables monetary transactions among peers on a blockchain network independent of a central authorizing agency. Ethereum smart contracts are programs that are deployed as decentralized applications, having the building blocks of the blockchain consensus protocol. This enables consumers to make agreements in a transparent and conflict-free environment. However, there exist some security vulnerabilities within these smart contracts that are a potential threat to the applications and their consumers and have shown in the past to cause huge financial losses. In this study, we review the existing literature and broadly classify the BT applications. As Ethereum smart contracts find their application mostly in e-commerce applications, we believe these are more commonly vulnerable to attacks. In these smart contracts, we mainly focus on identifying vulnerabilities that programmers and users of smart contracts must avoid. This paper aims at explaining eight vulnerabilities that are specific to the application level of BT by analyzing the past exploitation case scenarios of these security vulnerabilities. We also review some of the available tools and applications that detect these vulnerabilities in terms of their approach and effectiveness. We also investigated the availability of detection tools for identifying these security vulnerabilities and lack thereof to identify some of them

Ethereum Smart contracts use blockchain to transfer values among peers on networks without central agency. These programs are deployed on decentralized applications running on top of the blockchain consensus protocol to enable people to make agreements in a transparent and conflict-free environment. The security vulnerabilities within those smart contracts are a potential threat to the applications and have caused huge financial losses to their users. In this paper, we present a framework that combines static and dynamic analysis to detect Reentrancy vulnerabilities in Ethereum smart contracts. This framework generates an attacker contract based on the ABI specifications of smart contracts under test and analyzes the contract interaction to precisely report Reentrancy vulnerability. We conducted a preliminary evaluation of our proposed framework on 5 modified smart contracts from Etherscan and our framework was able to detect the Reentrancy vulnerability in all our modified contracts. Our framework analyzes smart contracts statically to identify potentially vulnerable functions and then uses dynamic analysis to precisely confirm Reentrancy vulnerability, thus achieving increased performance and reduced false positives., Comment: arXiv admin note: text overlap with arXiv:2105.02852

Blockchain technology (BT) Ethereum Smart Contracts allows programmable transactions that involve the transfer of monetary assets among peers on a BT network independent of a central authorizing agency. Ethereum Smart Contracts are programs that are deployed as decentralized applications, having the building blocks of the blockchain consensus protocol. This technology enables consumers to make agreements in a transparent and conflict-free environment. However, the security vulnerabilities within these smart contracts are a potential threat to the applications and their consumers and have shown in the past to cause huge financial losses. In this paper, we propose a framework that combines static and dynamic analysis to detect Denial of Service (DoS) vulnerability due to an unexpected revert in Ethereum Smart Contracts. Our framework, SmartScan, statically scans smart contracts under test (SCUTs) to identify patterns that are potentially vulnerable in these SCUTs and then uses dynamic analysis to precisely confirm their exploitability of the DoS-Unexpected Revert vulnerability, thus achieving increased performance and more precise results. We evaluated SmartScan on a set of 500 smart contracts collected from the Etherscan. Our approach shows an improvement in precision and recall when compared to available state-of-the-art techniques.

Deep neural networks (DNNs) show promise in image-based medical diagnosis, but cannot be fully trusted since their performance can be severely degraded by dataset shifts to which human perception remains invariant. If we can better understand the differences between human and machine perception, we can potentially characterize and mitigate this effect. We therefore propose a framework for comparing human and machine perception in medical diagnosis. The two are compared with respect to their sensitivity to the removal of clinically meaningful information, and to the regions of an image deemed most suspicious. Drawing inspiration from the natural image domain, we frame both comparisons in terms of perturbation robustness. The novelty of our framework is that separate analyses are performed for subgroups with clinically meaningful differences. We argue that this is necessary in order to avert Simpson's paradox and draw correct conclusions. We demonstrate our framework with a case study in breast cancer screening, and reveal significant differences between radiologists and DNNs. We compare the two with respect to their robustness to Gaussian low-pass filtering, performing a subgroup analysis on microcalcifications and soft tissue lesions. For microcalcifications, DNNs use a separate set of high frequency components than radiologists, some of which lie outside the image regions considered most suspicious by radiologists. These features run the risk of being spurious, but if not, could represent potential new biomarkers. For soft tissue lesions, the divergence between radiologists and DNNs is even starker, with DNNs relying heavily on spurious high frequency components ignored by radiologists. Importantly, this deviation in soft tissue lesions was only observable through subgroup analysis, which highlights the importance of incorporating medical domain knowledge into our comparison framework.

Crowdsourcing is a valuable approach for tracking objects in videos in a more scalable manner than possible with domain experts. However, existing frameworks do not produce high quality results with non-expert crowdworkers, especially for scenarios where objects split. To address this shortcoming, we introduce a crowdsourcing platform called CrowdMOT, and investigate two micro-task design decisions: (1) whether to decompose the task so that each worker is in charge of annotating all objects in a sub-segment of the video versus annotating a single object across the entire video, and (2) whether to show annotations from previous workers to the next individuals working on the task. We conduct experiments on a diversity of videos which show both familiar objects (aka - people) and unfamiliar objects (aka - cells). Our results highlight strategies for efficiently collecting higher quality annotations than observed when using strategies employed by today's state-of-art crowdsourcing system., Comment: CSCW 2020

Algorithmic methods for the explicit inversion of the indefinite double covering maps are proposed. These are based on either the Givens decomposition or the polar decomposition of the given matrix in the proper, indefinite orthogonal group with signature (p,q). As a by-product we establish that the preimage in the covering group, of a positive matrix in the proper, indefinite orthogonal group, can itself be chosen to be positive definite. Inversion amounts to solving a polynomial system in several variables. Our methods solve this system by either inspection, Groebner bases or by inverting the associated Lie algebra isomorphism and computing certain exponentials explicitly. The last method extends fully constructively for all (p,q) when used together with Givens decompositions. The remaining methods require details of the matrix form of the covering map, but then provide more information about the preimage. The techniques are illustrated for (p, q)in {(2,1), (2,2), (3,2), (4,1)}., Comment: 46 pages; some of the work was completed in 2016 in the first author's thesis. The balance was completed in 2018 and will form part of the third author's thesis

We assessed whether an adaptation of value-added analysis (VAA) can provide evidence on the relative effectiveness of interventions implemented in a large school district. We analyzed two datasets, one documenting interventions received by underperforming students, and one documenting interventions received by students in schools benefiting from discretionary funds to invest in specific programs. Results from the former dataset identified several interventions that appear to be more or less effective than the average intervention. Results from the second dataset were counterintuitive. We conclude that, under specific conditions, program VAA can provide evidence to help guide district decision-makers to identify outlier interventions and inform decisions about scaling up or disinvesting in such interventions, with the caveat that if those conditions are not met, the results could be misleading. [This is the online first version of an article published in "American Journal of Evaluation."]

Users of cloud computing are increasingly overwhelmed with the wide range of providers and services offered by each provider. As such, many users select cloud services based on description alone. An emerging alternative is to use a decision support system (DSS), which typically relies on gaining insights from observational data in order to assist a customer in making decisions regarding optimal deployment or redeployment of cloud applications. The primary activity of such systems is the generation of a prediction model (e.g. using machine learning), which requires a significantly large amount of training data. However, considering the varying architectures of applications, cloud providers, and cloud offerings, this activity is not sustainable as it incurs additional time and cost to collect training data and subsequently train the models. We overcome this through developing a Transfer Learning (TL) approach where the knowledge (in the form of the prediction model and associated data set) gained from running an application on a particular cloud infrastructure is transferred in order to substantially reduce the overhead of building new models for the performance of new applications and/or cloud infrastructures. In this paper, we present our approach and evaluate it through extensive experimentation involving three real world applications over two major public cloud providers, namely Amazon and Google. Our evaluation shows that our novel two-mode TL scheme increases overall efficiency with a factor of 60\% reduction in the time and cost of generating a new prediction model. We test this under a number of cross-application and cross-cloud scenarios.

We present a deep convolutional neural network for breast cancer screening exam classification, trained and evaluated on over 200,000 exams (over 1,000,000 images). Our network achieves an AUC of 0.895 in predicting whether there is a cancer in the breast, when tested on the screening population. We attribute the high accuracy of our model to a two-stage training procedure, which allows us to use a very high-capacity patch-level network to learn from pixel-level labels alongside a network learning from macroscopic breast-level labels. To validate our model, we conducted a reader study with 14 readers, each reading 720 screening mammogram exams, and find our model to be as accurate as experienced radiologists when presented with the same data. Finally, we show that a hybrid model, averaging probability of malignancy predicted by a radiologist with a prediction of our neural network, is more accurate than either of the two separately. To better understand our results, we conduct a thorough analysis of our network's performance on different subpopulations of the screening population, model design, training procedure, errors, and properties of its internal representations., Comment: MIDL 2019 [arXiv:1907.08612]

Background: The proliferation of cloud providers and provisioning levels has opened a space for cloud brokerage services. Brokers intermediate between cloud customers and providers to assist the customer in selecting the most suitable cloud service, helping to manage the dimensionality, heterogeneity, and uncertainty associated with cloud services. Objective: This paper identifies and classifies approaches to realise cloud brokerage. By doing so, this paper presents an understanding of the state of the art and a novel taxonomy to characterise cloud brokers. Method: We conducted a systematic literature survey to compile studies related to cloud brokerage and explore how cloud brokers are engineered. We analysed the studies from multiple perspectives, such as motivation, functionality, engineering approach, and evaluation methodology. Results: The survey resulted in a knowledge base of current proposals for realising cloud brokers. The survey identified surprising differences between the studies' implementations, with engineering efforts directed at combinations of market-based solutions, middlewares, toolkits, algorithms, semantic frameworks, and conceptual frameworks. Conclusion: Our comprehensive meta-analysis shows that cloud brokerage is still a formative field. There is no doubt that progress has been achieved in the field but considerable challenges remain to be addressed. This survey identifies such challenges and directions for future research.

Users of cloud computing are overwhelmed with choice, even within the services offered by one provider. As such, many users select cloud services based on description alone. We investigate the degree to which such strategy is optimal. In this quantitative study, we investigate the services of 2 of major IaaS providers. We use 2 representative applications to obtain longitudinal observations over 7 days of the week and over different times of the day, totalling over 14,000 executions. We give evidence of significant variations of performance offered within IaaS services, calling for brokers to use automated and adaptive decision making processes with means for incorporating expressive user constraints.

Decision making in cloud environments is quite challenging due to the diversity in service offerings and pricing models, especially considering that the cloud market is an incredibly fast moving one. In addition, there are no hard and fast rules, each customer has a specific set of constraints (e.g. budget) and application requirements (e.g. minimum computational resources). Machine learning can help address some of the complicated decisions by carrying out customer-specific analytics to determine the most suitable instance type(s) and the most opportune time for starting or migrating instances. We employ machine learning techniques to develop an adaptive deployment policy, providing an optimal match between the customer demands and the available cloud service offerings. We provide an experimental study based on extensive set of job executions over a major public cloud infrastructure., Comment: In the IEEE/IFIP Network Operations and Management Symposium (NOMS), April 2016

OPNET Modeler is used to simulate the architecture and to calculate the performance criteria (i.e. throughput, delay and data dropped) that slightly concerned in network estimation. It is concluded that our models shorten the time quite a bit for obtaining the performance measures of an end-to-end delay as well as throughput can be used as an effective tool for this purpose.