Papers by Alessandra Mileo
Applied Sciences
Advanced data augmentation techniques have demonstrated great success in deep learning algorithms... more Advanced data augmentation techniques have demonstrated great success in deep learning algorithms. Among these techniques, single-image-based data augmentation (SIBDA), in which a single image’s regions are randomly erased in different ways, has shown promising results. However, randomly erasing image regions in SIBDA can cause a loss of the key discriminating features, consequently misleading neural networks and lowering their performance. To alleviate this issue, in this paper, we propose the random slices mixing data augmentation (RSMDA) technique, in which slices of one image are placed onto another image to create a third image that enriches the diversity of the data. RSMDA also mixes the labels of the original images to create an augmented label for the new image to exploit label smoothing. Furthermore, we propose and investigate three strategies for RSMDA: (i) the vertical slices mixing strategy, (ii) the horizontal slices mixing strategy, and (iii) a random mix of both strat...
2022 IEEE International Conference on Big Data (Big Data)
Recent developments in in-situ monitoring and process control in Additive Manufacturing (AM), als... more Recent developments in in-situ monitoring and process control in Additive Manufacturing (AM), also known as 3D-printing, allows the collection of large amounts of emission data during the build process of the parts being manufactured. This data can be used as input into 3D and 2D representations of the 3D-printed parts. However the analysis and use, as well as the characterization of this data still remains a manual process. The aim of this paper is to propose an adaptive human-in-the-loop approach using Machine Learning techniques that automatically inspect and annotate the emissions data generated during the AM process. More specifically, this paper will look at two scenarios: firstly, using convolutional neural networks (CNNs) to automatically inspect and classify emission data collected by in-situ monitoring and secondly, applying Active Learning techniques to the developed classification model to construct a human-in-the-loop mechanism in order to accelerate the labeling process of the emission data. The CNN-based approach relies on transfer learning and fine-tuning, which makes the approach applicable to other industrial image patterns. The adaptive nature of the approach is enabled by uncertainty sampling strategy to automatic selection of samples to be presented to human experts for annotation.
arXiv (Cornell University), Jul 1, 2020
In recent years we have seen significant advances in the technology used to both publish and cons... more In recent years we have seen significant advances in the technology used to both publish and consume Linked Data. However, in order to support the next generation of ebusiness applications on top of interlinked machine readable data suitable forms of access control need to be put in place. Although a number of access control models and frameworks have been put forward, very little research has been conducted into the security implications associated with granting access to partial data or the correctness of the proposed access control mechanisms. Therefore the contributions of this paper are two fold: we propose a query rewriting algorithm which can be used to partially restrict access to SPARQL 1.1 queries and updates; and we demonstrate how a set of criteria, which was originally used to verify that an access control policy holds over different database states, can be adapted to verify the correctness of access control via query rewriting.
Paladyn, Journal of Behavioral Robotics, 2021
This article explores the rapidly advancing innovation to endow robots with social intelligence c... more This article explores the rapidly advancing innovation to endow robots with social intelligence capabilities in the form of multilingual and multimodal emotion recognition, and emotion-aware decision-making capabilities, for contextually appropriate robot behaviours and cooperative social human–robot interaction for the healthcare domain. The objective is to enable robots to become trustworthy and versatile social robots capable of having human-friendly and human assistive interactions, utilised to better assist human users’ needs by enabling the robot to sense, adapt, and respond appropriately to their requirements while taking into consideration their wider affective, motivational states, and behaviour. We propose an innovative approach to the difficult research challenge of endowing robots with social intelligence capabilities for human assistive interactions, going beyond the conventional robotic sense-think-act loop. We propose an architecture that addresses a wide range of soc...
Dealing with context is one of the most interesting and important problems faced in Artificial In... more Dealing with context is one of the most interesting and important problems faced in Artificial Intelligence (AI). Traditional AI applications often require to model, store, retrieve and reason about knowledge that holds within certain circumstances—the context. Without considering this contextual information, reasoning can easily run into problems such as: inconsistency, when considering knowledge in the wrong context; inefficiency, by considering knowledge irrelevant for a certain context; incompleteness, since an inference may depend on knowledge assumed to hold for a context but which is not explicitly stated. Contextual information is also relevant in knowledge representation and reasoning and it represents a strategic aspect to deal with inconsistency, ambiguity, uncertainty, knowledge base evolution, and commonsense reasoning, among others. In recent years, research in context-aware knowledge representation and reasoning became more relevant in the areas of Semantic Web and In...
Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence, 2019
The ability to enhance deep representations with prior knowledge is receiving a lot of attention ... more The ability to enhance deep representations with prior knowledge is receiving a lot of attention from the AI community as a key enabler to improve the way modern Artificial Neural Networks (ANN) learn. In this paper we introduce our approach to this task, which comprises of a knowledge extraction algorithm, a knowledge injection algorithm and a common intermediate knowledge representation as an alternative to traditional neural transfer. As a result of this research, we envisage a knowledge-enhanced ANN, which will be able to learn, characterise and reuse knowledge extracted from the learning process, thus enabling more robust architecture-agnostic neural transfer, greater explainability and further integration of neural and symbolic approaches to learning.
Web Reasoning and Rule Systems : 9th International Conference, RR 2015, Berlin, Germany, August 4-5, 2015, Proceedings
This book constitutes the refereed proceedings of the 9th International Conference on Web Reasoni... more This book constitutes the refereed proceedings of the 9th International Conference on Web Reasoning and Rule Systems, RR 2015, held in Berlin, Germany, in August 2015. The 5 full papers, 4 technical communications presented together with 4 invited talks were carefully reviewed and selected from 16 submissions. The scale and the heterogenous nature of web data poses many challenges, and turns basic tasks such as query answering and data transformations into complex reasoning problems. Rule-based systems have found many applications in this area. The RR conference welcomes original research from all areas of Web Reasoning and Rule Systems. Topics of particular interest are: answer set programming, complex events, datalog, description logics, event-condition-action rules, information extraction, and logic programming
Future Generation Computer Systems, 2021
The popularity of deep learning has increased tremendously in recent years due to its ability to ... more The popularity of deep learning has increased tremendously in recent years due to its ability to efficiently solve complex tasks in challenging areas such as computer vision and language processing. Despite this success, low-level neural activity reproduced by Deep Neural Networks (DNNs) generates extremely rich representations of the data. These representations are difficult to characterise and cannot be directly used to understand the decision process. In this paper we build upon our exploratory work where we introduced the concept of a co-activation graph and investigated the potential of graph analysis for explaining deep representations. The co-activation graph encodes statistical correlations between neurons' activation values and therefore helps to characterise the relationship between pairs of neurons in the hidden layers and output classes. To confirm the validity of our findings, our experimental evaluation is extended to consider datasets and models with different levels of complexity. For each of the considered datasets we explore the co-activation graph and use graph analysis to detect similar classes, find central nodes and use graph visualisation to better interpret the outcomes of the analysis. Our results show that graph analysis can reveal important insights into how DNNs work and enable partial explainability of deep learning models.
We present a probabilistic inductive logic programming framework which integrates non-monotonic r... more We present a probabilistic inductive logic programming framework which integrates non-monotonic reasoning, probabilistic inference and parameter learning. In contrast to traditional approaches to probabilistic Answer Set Programming (ASP), our framework imposes only comparatively little restrictions on probabilistic logic programs in particular, it allows for ASP as well as FOL syntax, and for precise as well as imprecise (interval valued) probabilities. User-configurable sampling and inference algorithms, which can be combined in a pipeline-like fashion, provide for general as well as specialized, more scalable approaches to uncertainty reasoning, allowing for adaptability with regard to different reasoning and learning tasks.
C-ASP: Continuous ASP-Based Reasoning over RDF Streams
Logic Programming and Nonmonotonic Reasoning, 2019
The ability to perform complex reasoning over data streams has recently become an important area ... more The ability to perform complex reasoning over data streams has recently become an important area of research in the Semantic Web community. Most of SPARQL-inspired engines have limitations in capturing sophisticated user requirements and dealing with complex reasoning tasks. To address these challenges, we propose and implement C-ASP, a reasoning system based on the Answer Set Programming (ASP) system Clingo and extended to handle continuous reasoning requests over RDF streams. We provide the syntax of the C-ASP language, as well as a set of examples in order to illustrate its expressive power. In addition, we present preliminary experimental results showing C-ASP performances.
Journal of Web Semantics, 2017
Enterprise Communication Systems are designed in such a way to maximise the efficiency of communi... more Enterprise Communication Systems are designed in such a way to maximise the efficiency of communication and collaboration within the enterprise. With users becoming mobile, the Internet of Things (IoT) can play a crucial role in this process, but is far from being seamlessly integrated into modern online communications. In this paper, we present a semantic infrastructure for gathering, integrating and reasoning upon heterogeneous, distributed and continuously changing data streams by means of semantic technologies and rule-based inference. Our solution exploits semantics to go beyond today's ad-hoc integration and processing of heterogeneous data sources for static and streaming data. It provides flexible and efficient processing techniques that can transform low-level data into high-level abstractions and actionable knowledge, bridging the gap between IoT and online Enterprise Communication Systems. We document the technologies used for acquisition and semantic enrichment of sensor data, continuous semantic query processing for integration and filtering, as well as stream reasoning for decision support. Our main contributions are the following, i) we define and deploy a semantic processing pipeline for IoT-enabled Communication Systems, which builds upon existing systems for semantic data acquisition, continuous query processing and stream reasoning, detailing the implementation of each component of our framework; ii) we present a rich semantic information model for representing and linking IoT data, social data and personal data in the Enterprise Communication scenario, by reusing and extending existing standard semantic models; iii) we define and develop an expressive stream reasoning component as part of our framework, based on continuous query processing and non-monotonic reasoning for semantic streams, iv) we conduct experiments to comparatively evaluate the performance of our data acquisition and semantic annotation layer based on OpenIoT, and the performance of our expressive reasoning layer in the scenario of Enterprise Communication.
A System for Probabilistic Inductive Answer Set Programming
Lecture Notes in Computer Science, 2015
We describe a prototypical software framework for probabilistic inductive logic programming which... more We describe a prototypical software framework for probabilistic inductive logic programming which supports the seamless combination of non-monotonic reasoning, probabilistic inference and parameter learning. While building upon existing as well as new approaches to probabilistic Answer Set Programming, our framework distinguishes itself from related works by placing virtually no restrictions on the annotation of knowledge with probabilities. User-configurable algorithms provide for general as well as specialized, scalable approaches to inference and parameter learning, allowing for adaptability with regard to complex reasoning and weight learning tasks.
Knowledge Representation and Reasoning for Sensory-rich Smart Environments to Support Independent Living
In recent years there has been growing interest in solutions for the delivery of clinical care fo... more In recent years there has been growing interest in solutions for the delivery of clinical care for the elderly, due to the large increase in aging population. Monitoring a patient in his home environment is necessary to ensure continuity of care in home settings, but, to be useful, this activity must not be too invasive for patients and a burden for caregivers. In this chapter we want to consider how knowledge representation and reasoning techniques can be used in sensory-rich environments , where data about the person and the environment conditions are collected through pervasive Wireless Sensor Networks (WSN), and expressive knowledge representation and reasoning techniques can be used to combine these data with external knowledge sources at a symbolic level to support caregivers in understanding patients' well being and in predicting possible evolutions of their health. A hierarchical logic-based model of health is able to combine data from different sources, sensor data, tes...
Sixth IEEE International Workshop on Policies for Distributed Systems and Networks (POLICY'05)
We describe a modified Grid architecture that allows the specification and enforcement connection... more We describe a modified Grid architecture that allows the specification and enforcement connection policies to Grid on Web Services. This is accomplished by interposing a policy enforcement engine between a calling application and the relative client stubs. Only service requests need be analized and filtered. Connection policies are conveniently expressed in the declarative policy specification language PPDL which allows expressing simple preferences and integrity constraints. PPDL policies are evaluated by translating them into a Logic Program with Ordered Disjunctions and calling the psmodels interpreter. This process is completely transparent to both client applications and Web/Grid services. There are clear advantages in having the connection logic expressed declaratively and externally to applications. * This work was supported by the Information Society Technologies programme of the European Commission, Future and Emerging Technologies under the IST-2001-37004 WASP project. Thanks also to J.Lobo, G.Gonzales and G.Gottlob for useful discussions on these topics. 1 The architecture presented here is intended for Web Services but it has been developed on top of the Globus Grid architecture. Therefore we will sometimes feel free to refer to them interchangeably 2 So far, however, we have considered only Java applications.
Lecture Notes in Computer Science, 2014
The proliferation of sensor devices and services along with the advances in event processing brin... more The proliferation of sensor devices and services along with the advances in event processing brings many new opportunities as well as challenges. It is now possible to provide, analyze and react upon realtime, complex events about physical or social environments. When existing event services do not provide such complex events directly, an event service composition maybe required. However, it is difficult to determine which event service candidates (or service compositions) best suit users' and applications' quality-of-service requirements. In this paper, we address this issue by first providing a quality-of-service aggregation schema for complex event service compositions and then developing a genetic algorithm to efficiently create optimal event service compositions.
Grid Service Selection with PPDL
Lecture Notes in Computer Science, 2004
StreamRule: A Nonmonotonic Stream Reasoning System for the Semantic Web
Web Reasoning and Rule Systems, 2013
Web Services
Lecture Notes in Computer Science, 2004
Uploads
Papers by Alessandra Mileo