Search results for "Machine Learning"
showing 10 items of 1464 documents
Graph Comparison and Artificial Models for Simulating Real Criminal Networks
2021
Network Science is an active research field, with numerous applications in areas like computer science, economics, or sociology. Criminal networks, in particular, possess specific topologies which allow them to exhibit strong resilience to disruption. Starting from a dataset related to meetings between members of a Mafia organization which operated in Sicily during 2000s, we here aim to create artificial models with similar properties. To this end, we use specific tools of Social Network Analysis, including network models (Barabási-Albert identified to be the most promising) and metrics which allow us to quantify the similarity between two networks. To the best of our knowledge, the DeltaCo…
Face Expression Recognition through Broken Symmetries
2008
Security systems, criminology, physical access control and man-machine interactions are examples of applications where recognition of human faces may be crucial. In the present paper a new signature, based on a measure of axial symmetry called DST, is proposed as a significant feature to analyze facial expressions. The measure of symmetry is an elaborate difference between the internal and external symmetry kernels of an object. The idea here is to use the evolution of the symmetry measure of a face over an ordered set of its sub-images. We claim that different evolutionary trends will represent different face expressions. The proposed signature has been tested on several face databases (ps…
Adversarial Machine Learning in e-Health: Attacking a Smart Prescription System
2022
Machine learning (ML) algorithms are the basis of many services we rely on in our everyday life. For this reason, a new research line has recently emerged with the aim of investigating how ML can be misled by adversarial examples. In this paper we address an e-health scenario in which an automatic system for prescriptions can be deceived by inputs forged to subvert the model's prediction. In particular, we present an algorithm capable of generating a precise sequence of moves that the adversary has to take in order to elude the automatic prescription service. Experimental analyses performed on a real dataset of patients' clinical records show that a minimal alteration of the clinical record…
Short-Term Sensory Data Prediction in Ambient Intelligence Scenarios
2014
Predicting data is a crucial ability for resource-constrained devices like the nodes of a Wireless Sensor Network. In the context of Ambient Intelligence scenarios, in particular, short-term sensory data prediction becomes a key enabler for more difficult tasks such as prolonging network lifetime, reducing the amount of communication required and improving user-environment interaction. In this chapter we propose a software module designed for clustered wireless sensor networks, able to predict various environmental quantities, namely temperature, humidity and light. The software module is supported by an ontology that describes the topology of the AmI scenario and the effects of the actuato…
Mimicking biological mechanisms for sensory information fusion
2013
Current Artificial Intelligence systems are bound to become increasingly interconnected to their surrounding environment in the view of the newly rising Ambient Intelligence (AmI) perspective. In this paper, we present a comprehensive AmI framework for performing fusion of raw data, perceived by sensors of different nature, in order to extract higher-level information according to a model structured so as to resemble the perceptual signal processing occurring in the human nervous system. Following the guidelines of the greater BICA challenge, we selected the specific task of user presence detection in a locality of the system as a representative application clarifying the potentialities of …
An Intelligent Empowering Agent (IEA) to Provide Easily Understood and Trusted Health Information Appropriate to the User Needs
2023
AbstractMost members of the public, including patients, usually obtain health information from Web searches using generic search engines, which is often overwhelming, too generic, and of poor quality. Although patients may be better informed, they are often none the wiser and not empowered to communicate with medical professionals so that their care is compatible with their needs, values, and best interests. Intelligent Empowering Agents (IEA) use AI to filter medical information and assist the user in the understanding of health information about specific complaints or health in general. We have designed and developed a prototype of an IEA that dialogues with the user in simple language, c…
Simulated Annealing Technique for Fast Learning of SOM Networks
2011
The Self-Organizing Map (SOM) is a popular unsupervised neural network able to provide effective clustering and data visualization for multidimensional input datasets. In this paper, we present an application of the simulated annealing procedure to the SOM learning algorithm with the aim to obtain a fast learning and better performances in terms of quantization error. The proposed learning algorithm is called Fast Learning Self-Organized Map, and it does not affect the easiness of the basic learning algorithm of the standard SOM. The proposed learning algorithm also improves the quality of resulting maps by providing better clustering quality and topology preservation of input multi-dimensi…
A Smart Assistant for Visual Recognition of Painted Scenes
2021
Nowadays, smart devices allow people to easily interact with the surrounding environment thanks to existing communication infrastructures, i.e., 3G/4G/5G or WiFi. In the context of a smart museum, data shared by visitors can be used to provide innovative services aimed to improve their cultural experience. In this paper, we consider as case study the painted wooden ceiling of the Sala Magna of Palazzo Chiaramonte in Palermo, Italy and we present an intelligent system that visitors can use to automatically get a description of the scenes they are interested in by simply pointing their smartphones to them. As compared to traditional applications, this system completely eliminates the need for…
Context-awareness for multi-sensor data fusion in smart environments
2016
Multi-sensor data fusion is extensively used to merge data collected by heterogeneous sensors deployed in smart environments. However, data coming from sensors are often noisy and inaccurate, and thus probabilistic techniques, such as Dynamic Bayesian Networks, are often adopted to explicitly model the noise and uncertainty of data. This work proposes to improve the accuracy of probabilistic inference systems by including context information, and proves the suitability of such an approach in the application scenario of user activity recognition in a smart home environment. However, the selection of the most convenient set of context information to be considered is not a trivial task. To thi…
A Federated Learning Approach for Distributed Human Activity Recognition
2022
In recent years, the widespread diffusion of smart pervasive devices able to provide AI-based services has encouraged research in the definition of new distributed learning paradigms. Federated Learning (FL) is one of the most recent approaches which allows devices to collaborate to train AI-based models, whereas guarantying privacy and lower communication costs. Although different studies on FL have been conducted, a general and modular architecture capable of performing well in different scenarios is still missing. Following this direction, this paper proposes a general FL framework whose validity is assessed by considering a distributed activity recognition scenario in which users' perso…