409 research outputs found
Sort by
Exploiting Process Algebras and BPM Techniques for Guaranteeing Success of Distributed Activities
Full text linkThe communications and collaborations among activities, pro-
cesses, or systems, in general, are the base of complex sys-
tems defined as distributed systems. Given the increasing
complexity of their structure, interactions, and functionali-
ties, many research areas are interested in providing mod-
elling techniques and verification capabilities to guarantee
their correctness and satisfaction of properties. In particular,
the formal methods community provides robust verification
techniques to prove system properties. However, most ap-
proaches rely on manually designed formal models, making
the analysis process challenging because it requires an expert
in the field. On the other hand, the BPM community pro-
vides a widely used graphical notation (i.e., BPMN) to design
internal behaviour and interactions of complex distributed
systems that can be enhanced with additional features (e.g.,
privacy technologies). Furthermore, BPM uses process min-
ing techniques to automatically discover these models from
events observation. However, verifying properties and ex-
pected behaviour, especially in collaborations, still needs a
solid methodology.
This thesis aims at exploiting the features of the formal meth-
ods and BPM communities to provide approaches that en-
able formal verification over distributed systems. In this con-
text, we propose two approaches. The modelling-based ap-
proach starts from BPMN models and produces process al-
gebra specifications to enable formal verification of system
properties, including privacy-related ones. The process mining-
based approach starts from logs observations to automati-
xv
cally generate process algebra specifications to enable veri-
fication capabilities
Type discipline for message-passing components in distributed systems
Full text linkComponent based software engineering (CBSE) is a method-
ology that aims to design and build software systems by
assembling together reusable and loosely coupled compo-
nents. Applying CBSE in a distributed setting is appealing
but challenging: distributed applications require different
remote components to interact following a well-defined
protocol. This thesis addresses a model for message passing
component-based systems where components are assembled
together with the protocol itself. Components can there-
fore be independent from the protocol, and can react to
messages in a flexible way. This thesis studies how types
can capture component behaviour and can enable checking
the compatibility with a protocol. In particular, this thesis
proposes two type languages for reactive components: the
first language excludes choice terms, whereas the second one
includes them. We show the correspondence of component
and type behaviours, which entails a progress property for
components
Empirical insights into strategic competition, productivity and resilience of the Italian entrepreneurial system
No full textThe globalization of economic activities and the acceleration
of technological change brought about profound changes in
the competitive arena where firms strive to succeed. Enter-
prises must promptly adapt to change to survive and build
a sustainable advantage over their competitors. The need for
effective solutions to strengthen the entrepreneurial system
and foster a dynamic economy has stimulated the academic
and institutional debate on the drivers of firms’ competitive-
ness, productivity, and resilience to sudden shocks. The ac-
knowledged systemic nature of the firm suggests looking be-
yond its boundaries, at the system of relationships the enter-
prise is embedded in, to discover the triggers of the firm’s
development.
OECD and European Union countries soon acknowledged
how valuable inter-firm connections are for firms’ strategic
development, especially when it comes to micro, small and
medium-sized enterprises. As a result, policymakers com-
mitted to defining a policy agenda to encourage the sponta-
neous emergence of formal network ties and support the de-
velopment of localized connections in territorial areas. The
first two chapters of this doctoral dissertation are devoted to
empirically investigating the triggers of firms’ competitive-
ness and productivity, focusing on the contribution to firms’
performance of formal network ties secured by a specific legal
regime and localized regional processes (i.e., local spillovers).
Chapter 1 assesses the dynamic impact of inter-firm network
agreements (introduced by the decree-law n. 5/2009 con-
verted into law n.33/2009) on firms’ performance. Our ap-
proach to causal inference allows us to estimate heterogeneity-robust dynamic effects overcoming the issues affecting two-
way fixed effects DiD estimates in settings with a staggered
treatment rollout. We find that firms participating in formal-
ized networks can reap lasting benefits that keep growing at
least until the third year of cooperation, thus improving their
revenues, value added and EBITDA. The benefits of formal-
ized networks are even stronger for the subsample of micro-
enterprises, especially when they engage with larger part-
ners. Moreover, inter-firm formal networks deliver higher
advantages when most members are in the same travel-to-
work area. Further insights into the consequences of co-location
are discussed in Chapter 2. The study reveals the existence of
spatial dependence between nearby firms’ productivity, which
is supposed to be driven by geographically bounded pro-
cesses. Using secondary data on Italian technology-intensive
manufacturing firms, we exploit spatial econometric models
to estimate productivity spillovers across firms. The work
brings together family firms and regional studies as it points
out whether spatial proximity to family firms is a source of
positive or negative externalities. Our findings confirm that
proximity to patenting firms is a source of positive external-
ities. As a second result, we observe that the family’s in-
volvement in the ownership and management positions has
an overall negative indirect effect on nearby firms’ productiv-
ity. However, when family firms are innovators, the adverse
indirect effect vanishes. The study points out the critical role
of innovation in fostering the development of dynamic and
fertile regional environments. It also highlights the impor-
tance of co-location for public policy initiatives designed to
promote economic growth at a local level.
Recently, the entrepreneurial system has been severely hit by
the unprecedented shock caused by the outbreak of the COVID-
19 pandemic. To curtail the health and socio-economic con-
sequences of the spread of coronavirus (SARS-CoV-2), gov-ernments issued several measures, including mobility lim-
itations and interventions to sustain employment (e.g., fur-
lough schemes). Chapter 3 contributes to developing a new
research line by analyzing how changes in mobility streams
following government restrictions and behavioral adjustments
impacted the number of excess deaths and employee furloughs
recorded in Italy after the pandemic outbreak. To disentangle
the causal effect of mobility restrictions on both dependent
variables, we exploited rainfall patterns across Italian admin-
istrative regions as a source of exogenous variation in human
mobility. We find that a contraction in mobility effectively
prevents the most severe consequences of the pandemic, as
it leads to a COVID-19 mortality reduction. However, it in-
creases the use of employee furloughs, exacerbating unem-
ployment risk. The Chapter builds on these findings to dis-
cuss return-to-work policies and prioritizing policies for ad-
ministering COVID-19 vaccines in the most advanced stage
of the vaccination campaign.
All Chapters focus on the Italian case. Notwithstanding, this
dissertation provides insights into widely addressed topics
that animate the institutional and academic debate on an in-
ternational scale
Model Predictive Control for Legged Robots
Full text linkOptimal planning is essential when it comes to autonomy in
legged locomotion. In the last few decades, different optim-
ization techniques have been presented to design a legged lo-
comotion framework, such as Trajectory Optimization (TO)
and Model Predictive Control (MPC). The choice of a dy-
namic model utilized while synthesizing these planners plays
a pivotal role because the chosen model defines the accuracy
of the planning and also becomes a deciding factor for the
computational cost of these techniques. In the first part of
this thesis, we propose a closed-loop validation procedure for
the Single Rigid Body Dynamics (SRBD) model and its vari-
ants used for optimal planning. Thereafter, we introduce a
Linear Time-Varying (LTV) based TO for legged locomotion,
followed by the simulation results and discussion on its lim-
itations in re-planning.
Re-planning in legged locomotion is crucial to track the de-
sired user velocity while adapting to the terrain and reject-
ing external disturbances. In the second part of this thesis,
we propose and test in experiments a real-time Nonlinear
Model Predictive Control (NMPC) tailored to a legged robot
to achieve dynamic locomotion on various terrains. We in-
troduce a novel mobility-based criterion to define an NMPC
cost that enhances the locomotion of quadruped robots while
maximizing leg mobility and improving adaptation to the ter-
rain features. The NMPC is based on the Real-Time Iteration
(RTI) scheme that allows us to re-plan online at 25 Hz with a
prediction horizon of 2 seconds. In simulations, the NMPC is
tested to traverse a set of pallets of different sizes, walk into a
V-shaped chimney, and locomote over rough terrain. In real
experiments, we demonstrate the effectiveness of our NMPC
with the mobility feature that allowed IIT’s 87 kg quadruped
robot HyQ to achieve an omni-directional walk on flat terrain,
traverse a static pallet, and adapt to a repositioned pallet dur-
ing a walk.In the final part of this thesis, we present the extension of
the NMPC with other dynamic gaits, i.e., trot and pace.
We also introduce an Optimization-Based Reference Gener-
ator (ORG) that computes dynamically feasible trajectories
for the state and control input based on the Linear Inver-
ted Pendulum (LIP) model-based optimization and Quad-
ratic Programming (QP) based mapping. These feasible tra-
jectories are passed to the NMPC to cope with the disturb-
ances while following the user-defined trajectories with the
dynamic gaits. We show the effectiveness of this two-stage
optimization scheme in simulations and experiments per-
formed on the AlienGo robot to trot in a straight line and
to recover from the external disturbances while trotting. We
also compare the performance of the two-stage scheme with
respect to a traditional heuristic reference generator in an ex-
periment
Neural signatures of auditory statistics: a window into auditory computations and their interactions with other modalities
No full textThe auditory system processes information at high temporal
resolutions, extracting fine-grained details from complex
sounds. However, this ability comes at a cost as the acoustic
information often exceeds memory storage capacity. To keep
track of sound changes occurring over several seconds, the
auditory system abstracts local features into compact
representations (summary statistics). This thesis addresses
three questions: (i) whether it is possible to distinguish from
neural activity the processing of local features or summary
statistics; (ii) whether the brain is endowed with distinct
structures for computations based on local features or
summary statistics; (iii) whether these basic computations
are affected by other sensory modalities.
First, we designed a protocol for the EEG. Participants were
exposed to streams comprising triplets of synthetic sound
excerpts. Two sounds were identical, while the third could
vary for its local features or summary statistics. We
presented sounds of different durations to manipulate the
similarity of statistics measured from the repeated and novel
sounds. Results showed that local details and summary
statistics are processed automatically and encoded by
different neural oscillatory profiles. Second, we collected
MEG data with the same protocol and performed source
reconstruction of the evoked response to the novel sounds.
This analysis revealed functional cortical specializations and
hemispheric asymmetries for the processing of computations
occurring at high or low temporal resolutions. Third, we
tested three groups of individuals, congenitally (CB), late-
onset blinds (LB), and sighted controls (SC) in two
behavioral experiments. One benefitted from the processing
of local features, the other from summary statistics. CB
performed as SC in both tasks, showing that both
computations can develop independently from vision.
Conversely, LB’s performance was impaired when relying on local features, with no alterations in summary statistics
processing. These findings suggest an audiovisual interplay
selectively for processing auditory details, which emerges
only in late development. Overall, these findings
demonstrate that the auditory system utilizes distinct neural
processes and dedicated brain structures to encode local
features and summary statistics of sound and emphasize the
role of visual experience in the processing of local features.
By unraveling these fundamental aspects of auditory
perception, this thesis expands our knowledge in the context
of auditory cognition and its complex interplay with other
sensory modalities
Beyond Dichotomy: Exploring the Intersection of Semantic and Sensory Information in Abstract and Concrete Words Formation and Representation. Insights from Superordinate words.
No full textThis thesis is focused on the description of conceptualization
mechanisms that allow to create unified and shared representations
of percepts. Moreover, the differences and similarities between
abstract and concrete word semantic and conceptual representations
are analysed. In particular, the definitions of abstraction and
abstractness are evaluated in order to disentangle them. Reviewing
the available literature on the topic, from the point of view of the
different disciplines that have tackled the matter, from philosophy to
cognitive sciences, unresolved issues are reported and put forward,
advocating the need to go beyond the classical dichotomic
subdivision of abstract and concrete words. The authors put forward
the need to take into account the architecture of semantic
representations when dealing with studies on words and concepts
processing. The aim of the study is to assess the importance of
sensory information and semantic architecture in conceptual
representation, particularly focusing on these questions:
What is the role of sensory information in concept formation and
retrieval? Does knowledge depend on modality-dependent
information, or is it organized in a more abstract semantic manner?
Is the presence (or lack) of sensory information (abstractness) or the
different semantic architecture (abstraction) that drives the different
behavioral and neural responses to concrete and abstract concepts?
We hypothesize that abstract and concrete concepts may differ on the
level of abstraction needed to process them. Moreover, in order to
disentangle the different contributions of sensory information and
semantic architecture, we included in the design superordinate
concepts, which are linked to sensory information but are
characterized by more general and less detailed semantic
representation.180 balanced stimuli (60 concrete, 60 abstracts 60 superordinate)
were selected and evaluated by 46 Italian native speakers with a 5-
point-Likert-scale on concreteness, abstractness, familiarity, and
generalizability. The same task was administered to 327 English
native speakers to assess interlinguistic agreement in the evaluation.
99 participants were asked to produce a maximum of ten features to
describe each word. These features were then segmented and
lemmatized and were used to evaluate the semantic richness of the
stimuli words (Relevance and Pointwise Mutual Information).
51 balanced stimuli (17 concrete, 17 abstract, and 17 superordinate)
were selected for the EEG study. The stimuli were balanced both for
length and frequency. Six Italian native speakers from all over Italy,
three males and three females recorded the stimuli. Recordings of the
stimuli were balanced for RMS and length. 20 Italian native speakers
took part in the EEG study. They were instructed to listen to the
words and think carefully about their meanings. As attention check,
they were asked for 10% of the trials, in a randomized order, to
evaluate the semantic similarity of the words heard and further
words which were not part of the database. The electrophysiological
data from the attention checks were then discarded and not analysed.
The sub-sample of 51 stimuli was evaluated on concreteness,
abstractness, familiarity, and generalizability by 18 blind
participants. The aim of this study was to evaluate whether lack of
sensory information experience lead to differences in the evaluation
of concrete, abstract and superordinate stimuli and see whether the
abstraction continuum hypothesized depended on sensory
information contribution and was then disrupted or significantly
different in the blind population. Behavioral results showed a
continuum from concrete, characterized by higher values of sensory
information and semantic architecture to superordinate to abstract
concepts, with the lowest values of sensory information and sematic
richness. ERPs differed significantly at the latencies 250-350ms and 650-700ms, with concrete concepts eliciting greater responses than
both abstract and superordinate concepts. Despite being grounded
in sensory information, EEG response to superordinate categories
was indistinguishable from abstract concepts, while both were
significantly different from concrete concepts. These results
highlight the importance of the semantic architecture, advocating a
redefinition of abstract and concrete concepts that encompass the
traditional dichotomy of sensory/non-sensory grounded concepts
Essays on the Economics of Labor Markets and Retirement Policies
Full text linkThis dissertation explores three distinct yet relevant aspects of la-
bor markets, shedding new light the micro- and macroeconomic
mechanisms behind them. It comprises three independent essays.
In the first chapter, I explore a novel mechanism through which
firms can provide value to their employees: reducing on-the-job
search frictions. I build a structural search model where the rate
of job offers depends on the current employer. Workers thus value
the firms’ contribution to accelerating their ascent on the job ladder.
Using a reduced-form approach, I demonstrate the existence of this
compensating differential and its payoff in terms of future earnings.
Finally, I structurally estimate the model, showing a precise fit with
the data.
The second chapter offers new evidence of the heterogeneous ef-
fects on firm productivity distribution caused by a labor market re-
form aimed at enhancing labor flexibility, which indirectly reduced
labor costs. Specifically, we show that this decrease in labor costs—
attributable to the unique features of Italian collective bargaining
institutions—suppresses total factor productivity (TFP) among al-
ready unproductive firms while increasing it for the most produc-
tive ones. We argue that this effect is driven by negative selection
at the bottom of the distribution and construct a model that ratio-
nalizes this mechanism and provides welfare implications.
The third chapter uses an overlapping generation model to study
the implications on optimal taxation of the government’s use of
a credible set of social security instruments. We reveal that these
instruments introduce new redistributive motives and crowd out
others in the context of a standard Ramsey problem. We calibrate
the model using data from three different economies, showing that
current retirement benefits exceed their optimal level and that the
implementation of funded social security schemes is desirable.
The dissertation contributes to various branches of labor economics
and macro-public finance literature: i. it investigates a brand new compensating differential channel for high-skilled workers that ex-
plains a significant component of employees’ transitions behavior;
ii. it presents new empirical and theoretical evidence on the hetero-
geneous effects of labor market reforms on productivity; iii. it char-
acterizes optimal distortionary labor and capital taxation for the
first time in the context of a rich set of social security instruments,
bridging the gap between social security and traditional Ramsey
policy instruments
Neural plasticity induced by different degrees of perturbation in auditory and visual sensory systems
Full text linkVisual or auditory sensory deprivation represents a key model for studying experience-dependent plasticity. Different types of deprivations (congenital-late, temporary-permanent, peripheral-central)are characterized by a certain degree of perturbation of the typical sensory experience and can be represented in a three-dimensional space
showing the distance from the typical experience. The dimensions are:(i) when the deprivation occurs, (ii) how long the deprivation lasts, and (iii) where is the barrier that causes the deprivation. Each dimension can
be responsible for a low, medium, or high degree of perturbation. In thisdissertation, visual and auditory deprivation models are employed to investigate unisensory and multisensory neural plasticity. The first
study (low degree of perturbation) aimed to unveil whether short-term monocular deprivation in the adult brain can induce neural plasticity beyond the visual system. The second study (medium degree of perturbation), using the model of temporary deprivation, assessed whether neural tracking of speech envelope could develop even in the
absence of auditory stimulation from birth. Finally, the third study (high degree of perturbation) investigated how cerebral visual impairment affects visuospatial processing. Neural oscillations were used as windows to investigate plasticity mechanisms; time-frequency analysis
was employed when short stimuli were presented, and neural tracking when the stimuli were continuous. Results revealed that even a low degree of sensory perturbation induces plasticity that extends beyond the deprived modality (study 1); altered neural tracking develops following a medium degree perturbation (study 2); a high degree of
perturbation has a widespread impact on neural activity (study 3). These results strengthen evidence of the pivotal role of sensory experience revealing multifaced aspects of experience-dependent plasticity.
Modeling the degree of perturbation could be a helpful perspective for a deeper understanding of how neural dynamics are affected by different types of deprivation and for shedding light on ranges of flexibility in neural processing with potential clinical implications
Invariant Set-based Methods for the Computation of Input and Disturbance Sets
Full text linkThis dissertation presents new methods to synthesize disturbance sets and input constraints set for constrained linear time-invariant systems. Broadly, we formulate and solve optimization problems that (a) compute disturbance sets such that the reachable set of outputs approximates an assigned set, and (b) compute input constraint sets guaranteeing the stabilizability of a given set of initial conditions. The proposed methods find application in the synthesis and analysis of several control schemes such as decentralized control, reduced-order control, etc., as well as in practical system design problems such as actuator selection, etc.
The key tools supporting the develpment of the aforementioned methods are Robust Positive Invariant (RPI) sets. In particular, the problems that we formulate are such that they co-synthesize disturbance/input constraint sets along with the associated RPI sets. This requires embedding existing techniques to compute RPI sets within an optimization problem framework, that we facilitate by developing new results related to properties of RPI sets, polytope representations, inclusion encoding techniques, etc.
In order to solve the resulting optimization problems, we develop specialized structure-exploiting solvers that we numerically demonstrate to outperform conventional solution methods. We also demonstrate several applications of the methods we propose for control design. Finally, we extend the methods to tackle data-driven control synthesis problems in an identification-for-control framework
Advances in macroeconometrics: (interpretable) machine learning and high-frequency data for forecasting and structural analysis
No full textForecasting and modelling techniques for structural analy-
sis have changed through the years to cope with the com-
plexity of macroeconomic systems. Recent results show evi-
dence that non-parametric models such as machine learning
are helping with the prediction of macroeconomic variables.
On the other side, high-frequency information is widely used
to provide a new source of information for structural analy-
sis. This thesis contributes to all these aspects by proposing
innovative approaches for forecasting macroeconomic indi-
cators and providing an alternative way to make structural
analysis. We first exploit the ability of an ensemble learning
model combining long-short-term memory neural network
(LSTM) and dynamic factor model (DFM) to detect nonlin-
earities in the US GDP forecast. We also provide an inter-
pretable methodological framework that uses Shapley values
to generalize the data-generating process learned by neural
networks and applies it to predict inflation levels. The result-
ing polynomial relations between the variables provide pol-
icymakers with valuable insights on the potential nonlinear
relations between the evolution of future price levels and eco-
nomic activity. In addition, we propose a new identification
method for Structural Vector Autoregressive (SVAR) models
based on nowcasted (high-frequency) macroeconomic data