Search

Explainable Artificial Intelligence (XAI) aims to provide insights into the decision-making process of AI models, allowing users to understand their results beyond their decisions. A significant goal of XAI is to improve the performance of AI models by providing explanations for their decision-making processes. However, most XAI literature focuses on how to explain an AI system, while less attention has been given to how XAI methods can be exploited to improve an AI system. In this work, a set of well-known XAI methods typically used with Machine Learning (ML) classification tasks are investigated to verify if they can be exploited, not just to provide explanations but also to improve the performance of the model itself. To this aim, two strategies to use the explanation to improve a classification system are reported and empirically evaluated on three datasets: Fashion-MNIST, CIFAR10, and STL10. Results suggest that explanations built by Integrated Gradients highlight input features that can be effectively used to improve classification performance., This work has been accepted to be presented to The 1st World Conference on eXplainable Artificial Intelligence (xAI 2023), July 26-28, 2023 - Lisboa, Portugal

A systematic review on machine-learning strategies for improving generalizability (cross-subjects and cross-sessions) electroencephalography (EEG) based in emotion classification was realized. In this context, the non-stationarity of EEG signals is a critical issue and can lead to the Dataset Shift problem. Several architectures and methods have been proposed to address this issue, mainly based on transfer learning methods. 418 papers were retrieved from the Scopus, IEEE Xplore and PubMed databases through a search query focusing on modern machine learning techniques for generalization in EEG-based emotion assessment. Among these papers, 75 were found eligible based on their relevance to the problem. Studies lacking a specific cross-subject and cross-session validation strategy and making use of other biosignals as support were excluded. On the basis of the selected papers' analysis, a taxonomy of the studies employing Machine Learning (ML) methods was proposed, together with a brief discussion on the different ML approaches involved. The studies with the best results in terms of average classification accuracy were identified, supporting that transfer learning methods seem to perform better than other approaches. A discussion is proposed on the impact of (i) the emotion theoretical models and (ii) psychological screening of the experimental sample on the classifier performances., under review

An interesting case of the well-known Dataset Shift Problem is the classification of Electroencephalogram (EEG) signals in the context of Brain-Computer Interface (BCI). The non-stationarity of EEG signals can lead to poor generalisation performance in BCI classification systems used in different sessions, also from the same subject. In this paper, we start from the hypothesis that the Dataset Shift problem can be alleviated by exploiting suitable eXplainable Artificial Intelligence (XAI) methods to locate and transform the relevant characteristics of the input for the goal of classification. In particular, we focus on an experimental analysis of explanations produced by several XAI methods on an ML system trained on a typical EEG dataset for emotion recognition. Results show that many relevant components found by XAI methods are shared across the sessions and can be used to build a system able to generalise better. However, relevant components of the input signal also appear to be highly dependent on the input itself., Accepted to be presented at XAI.it 2022 - Italian Workshop on Explainable Artificial Intelligence

In recent years, Artificial Neural Networks (ANNs) have been introduced in Structural Health Monitoring (SHM) systems. A semi-supervised method with a data-driven approach allows the ANN training on data acquired from an undamaged structural condition to detect structural damages. In standard approaches, after the training stage, a decision rule is manually defined to detect anomalous data. However, this process could be made automatic using machine learning methods, whom performances are maximised using hyperparameter optimization techniques. The paper proposes a semi-supervised method with a data-driven approach to detect structural anomalies. The methodology consists of: (i) a Variational Autoencoder (VAE) to approximate undamaged data distribution and (ii) a One-Class Support Vector Machine (OC-SVM) to discriminate different health conditions using damage sensitive features extracted from VAE's signal reconstruction. The method is applied to a scale steel structure that was tested in nine damage's scenarios by IASC-ASCE Structural Health Monitoring Task Group.

Over the last few years, we have witnessed the availability of an increasing data generated from non-Euclidean domains, which are usually represented as graphs with complex relationships, and Graph Neural Networks (GNN) have gained a high interest because of their potential in processing graph-structured data. In particular, there is a strong interest in exploring the possibilities in performing convolution on graphs using an extension of the GNN architecture, generally referred to as Graph Convolutional Neural Networks (ConvGNN). Convolution on graphs has been achieved mainly in two forms: spectral and spatial convolutions. Due to the higher flexibility in exploring and exploiting the graph structure of data, there is recently an increasing interest in investigating the possibilities that the spatial approach can offer. The idea of finding a way to adapt the network behaviour to the inputs they process to maximize the total performances has aroused much interest in the neural networks literature over the years. This paper presents a novel method to adapt the behaviour of a ConvGNN to the input proposing a method to perform spatial convolution on graphs using input-specific filters, which are dynamically generated from nodes feature vectors. The experimental assessment confirms the capabilities of the proposed approach, which achieves satisfying results using a low number of filters.

Controlling the body – given its huge number of degrees of freedom – poses severe computational challenges. Mounting evidence suggests that the brain alleviates this problem by exploiting “synergies”, or patterns of muscle activities (and/or movement dynamics and kinematics) that can be combined to control action, rather than controlling individual muscles of joints [1–10]. D’Ausilio et al. [11] explain how this view of motor organization based on synergies can profoundly change the way we interpret studies of action recognition in humans and monkeys, and in particular the controversy on the “granularity” of the mirror neuron system (MNs): whether it encodes either (lower) kinematic aspects of movements, or (higher) goal representations, or both but at different hierarchical levels [12]. Here we offer a complementary, computational perspective on the role of synergies for action recognition and the MNs. In computational modeling and robotics, it is widely assumed that a control scheme using synergies simplifies movement planning and execution. This scheme permits to use elemental behaviors or primitives as “building blocks” to be composed (e.g., combined linearly, sequenced) to produce more complex behaviors, thus controlling relatively few degrees of freedom [13–15]. Do synergies yield equivalent benefits for action recognition? To answer this question from a computational viewpoint, we frame the concept of synergies within generative architectures of action execution and recognition [16–20]. According to two leading theories of motor control, optimal feedback control [21] and active inference [22], the motor system can be conceptualized as a (hierarchical) generative model, which encodes a (probabilistic) mapping between “task goals” specified at a higher level (e.g., grasping a cup) and states of the “plant” to be controlled (i.e., (Less)

The computational model presented here, Grasping Affordances (GA) model, provides a precise explication of the notion of affordance in the context of grasping actions carried out by monkeys. This explication is consistent with both direct perception theories and neuroscientific models of monkey brains, insofar as the identification of grasping affordances requires, according to this model, neither object recognition processes nor access to semantic memory. Nevertheless, this model posits a cascade of complicated computational processes, in the way of visuo-motor transformations, which suggest the advisability of qualifying and re-interpreting the claim that (grasping) affordances are directly available to an acting biological system. This re-interpretation undermines the alleged alternative between direct and indirect perception theories, to the extent that substantive visuo-motor transformations have to be posited in order to identify grasping affordances

Target behaviours can be achieved by finding suitable parameters for Continuous Time Recurrent Neural Networks (CTRNNs) used as agent control systems. Differential Evolution (DE) has been deployed to search parameter space of CTRNNs and overcome granularity, boundedness and blocking limitations. In this paper we provide initial support for DE in the context of two sample learning problems.

We propose a new deflationary interpretation for the functional role of mirror neurons discovered by Rizzolatti and colleagues into the macaque’s F5 motor area. Several functional interpretations of the mirror activity have been proposed, emphasizing their role in action understanding and representing, language evolution, or mind-reading abilities. However, according to the interpretation presented here, full understanding of agent goals and intentional action are related to mirror neuron activity only insofar as the process of understanding an intentional action necessarily involves the capability of assigning a “structural description” to the action. In particular, we argue that mirror activity is involved in both recognizing action structural features and associating these features to motor commands. Our functional interpretation includes an anticipatory mechanism, enabling one to verify whether the actual, suitably coded visual input, matches an expected visual input computed on the basis of a motor command sequence. This mechanism is involved in both action recognition and action execution control. We believe that this mirror neuron model could fit biological data better than the Oztop and Arbib model. Starting from this interpretation, we propose a biologically inspired visuo-motor control model merging basic elements of the Oztop and Arbib model with an expected perception mechanism. This model has been formalized within an algorithmic stance however actual implementation is in progress.

Answer validation is the ability to automatically judge the relevance of a candidate answer with respect to a given question. This paper investigates answer validation following a data-driven approach that considers textual passages (i.e. snippets) retrieved from the Web as the main source of information. Snippets are then analyzed in order to maximize the density of relevant keywords belonging both to the question and to the answer. Results obtained on a corpus of human-judged factoid question-answer pairs submitted by participants at TREC-2001 show a satisfactory degree of success rate (i.e. 86\%). In addition, the efficiency of the methodology (documents are not downloaded) makes the approach suitable to be integrated as a module in the architecture of a question answering system.

{Answer Validation is an emerging topic in Question Answering, where open domain systems are often required to rank huge amounts of candidate answers. We present a novel approach to answer validation based on the intuition that the amount of implicit knowledge which connects an answer to a question can be estimated by exploiting the redundancy of Web information. Two techniques are considered in this paper: a statistical approach, which uses the Web to obtain a large amount of pages, and a content-based approach, which analyses text snippets retrieved by the search engine. Both the approaches do not require to download the documents. Experiments carried out on the TREC-2001 judged-answer collection show that a combination of the two approaches achieves a high level of performance (i.e. about 88% success rate). The simplicity and the efficiency of these Web-based techniques make them suitable to be used as a module in Question Answering systems.

This paper presents a new version of the DIOGENE Question Answering (QA) system developed at ITC-Irst. With respect to our first participation to the TREC QA main task (TREC-2001), the system presents both improvements and extensions. On one hand, significant improvements rely on the substitution of basic components (e.g. the search engine and the tool in charge of the named entities recognition) with new modules that enhanced the overall system’s performance. On the other hand, an effective extension of DIOGENE is represented by the introduction of a module for the automatic assessment of the candidate answers ’ quality. All the variations with respect to the first version of the system as well as the results obtained at the TREC-2002 QA main task are presented and discussed in the paper.

This paper presents the DIOGENE question/answering system developed at ITCIrst. The system is based on a rather standard architecture which includes three components for question processing, search and answer extraction. Linguistic processing strongly relies on MULTIWORDNET, an extended version of the English WORDNET. The system has been designed to address two promising directions: multilingual question/answering and question/answering on the Web. The results obtained in the TREC-10 main task will be presented and discussed.

In this paper a Boolean neural networks which is able to learn and to control temporal sequences of simple behaviors (called primitive) is prsented. This ability is based on a specific organization of subnetworks called assemblies. We discuss here a novel model of assembly.

Here, we discuss a Boolean neural networks which is able to learn and to control temporal sequences of simple behaviors (called primitive). This property is achieved by imposing a specific neural network organization. In particular, in this paper we underline that by this specific organization a Boolean neural network can be considered as ruled based system learning new rules by examples.

To train a Boolean neural network to comment on the progress of the task it is controlling, we have implemented in the network an Hebbian rule with the help of the node assemblies. Actually, our network can comment also on its communications with us without any danger to enter in a loop-like endeless activity