ddiruscio/logseq
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Auto saved by Logseq

Browse Source
This commit is contained in:
Davide Di Ruscio
2025-06-02 17:15:13 +02:00
commit d9ea3d552d
6279 changed files with 711145 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
pages/ReadingNotes/PRIN2021/ApplicationofAIandMLinIoTPdf - Green Annotations (17122020, 143313).md
+18
View File
@@ -0,0 +1,18 @@
* Mdnotes File Name: [[ApplicationofAIandMLinIoTPdf]]
# Green Annotations (17/12/2020, 14:33:13)
> "IoT with AI and ML in industries has the potential to transform their outputs and aid them to yield better results" ([ :13](zotero://open-pdf/library/items/RXAEHVFX?page=13))
> "he industries may also use connected tools and machinery that eliminate the chances of errors made while setting the parameters manually." ([ :13](zotero://open-pdf/library/items/RXAEHVFX?page=13))
> "In the future, it will be nearly impossible to find IoT systems that don't utilize AI services" ([ :18](zotero://open-pdf/library/items/RXAEHVFX?page=18))
> "the future of IoT is AI" ([ :18](zotero://open-pdf/library/items/RXAEHVFX?page=18))
> "he main purpose is to make life easier by working smarter and not harder." ([ :18](zotero://open-pdf/library/items/RXAEHVFX?page=18))
> "t is expected that in the future, we will have devices that will make our lives more convenient than it is today by performing our tasks timely, even before we think of doing it, and give us real-time data and insights about every aspect of our livesbe it personal, professional or social." ([ :18](zotero://open-pdf/library/items/RXAEHVFX?page=18))
> "automation which will bring out the true essence of IoT" ([ :18](zotero://open-pdf/library/items/RXAEHVFX?page=18))
pages/ReadingNotes/PRIN2021/cioffiArtificialIntelligenceMachine2020 - Green Annotations (18122020, 163502)Table 5___ Main areas in sustainable manufacturing___ (Cioffi et al 2020507)MAYBE.md
+8
View File
@@ -0,0 +1,8 @@
* Mdnotes File Name: [[cioffiArtificialIntelligenceMachine2020]]
# Green Annotations (18/12/2020, 16:35:02)
> "Table 5. Main areas in sustainable manufacturing." ([Cioffi et al 2020:507](zotero://open-pdf/library/items/YUS586FA?page=16))
MAYBE USEFUL FOR THE OBJECTIVES ([note on p.507](zotero://open-pdf/library/items/YUS586FA?page=16))
pages/ReadingNotes/PRIN2021/felfernigOverviewRecommenderSystems2019 - Yellow Annotations (17122020, 170517).md
+42
View File
@@ -0,0 +1,42 @@
* Mdnotes File Name: [[felfernigOverviewRecommenderSystems2019]]
# Yellow Annotations (17/12/2020, 17:05:17)
> "any system that guides a user in a personalized way to interesting or useful objects in a large space of possible options or that produces such objects as output" ([Felfernig et al 2019:285](zotero://open-pdf/library/items/PZH6FFNA?page=1))
> "collaborative filtering and content-based filtering. Collaborative Filtering (Konstan et al. 1997) is using the opinio" ([Felfernig et al 2019:286](zotero://open-pdf/library/items/PZH6FFNA?page=2))
> "IoT workflow development" ([Felfernig et al 2019:286](zotero://open-pdf/library/items/PZH6FFNA?page=2))
> "ecommendation of app" ([Felfernig et al 2019:286](zotero://open-pdf/library/items/PZH6FFNA?page=2))
> "domain-specific scenarios such as food recommendation (" ([Felfernig et al 2019:286](zotero://open-pdf/library/items/PZH6FFNA?page=2))
> "recommender engine can be included to assist users in the configuration of the gateway" ([Felfernig et al 2019:287](zotero://open-pdf/library/items/PZH6FFNA?page=3))
> "in recommending useful applications based on given gateway settings and user interaction protocols." ([Felfernig et al 2019:287](zotero://open-pdf/library/items/PZH6FFNA?page=3))
> "In the context of wildlife animal monitoring, measuring devices and data collection units (typically drones) have to be selected and parametrized in such a way that the observation area is completely covered" ([Felfernig et al 2019:287](zotero://open-pdf/library/items/PZH6FFNA?page=3))
> "In the context of smart homes, recommendation technologies improve the overall applicability of the installed equipment and can also help to optimize the usage of the available resources (e.g., minimizing power consumption)." ([Felfernig et al 2019:287](zotero://open-pdf/library/items/PZH6FFNA?page=3))
> "recommender systems can help the spectators to determine the current geographical location of certain athletes. This further results in recommended sites at which the athlete can be seen and cheered." ([Felfernig et al 2019:287](zotero://open-pdf/library/items/PZH6FFNA?page=3))
> "how recommenders can be applied in IoT scenarios and to propose new recommendation approaches for the IoT domain." ([Felfernig et al 2019:288](zotero://open-pdf/library/items/PZH6FFNA?page=4))
> "existing applications of recommendation technologies in the IoT." ([Felfernig et al 2019:288](zotero://open-pdf/library/items/PZH6FFNA?page=4))
> "IoT infrastructure, a recommender system does not have to only rely on the preferences of the user but can take into account further information sources." ([Felfernig et al 2019:288](zotero://open-pdf/library/items/PZH6FFNA?page=4))
> "Alex needs to receive some app, device or communication protocol (BLE, zigbee, etc.) recommendations according to the overall settings on the gateway." ([Felfernig et al 2019:289](zotero://open-pdf/library/items/PZH6FFNA?page=5))
> "support for choosing the sensors and configuring the system properly." ([Felfernig et al 2019:289](zotero://open-pdf/library/items/PZH6FFNA?page=5))
> "SEQREQ: Sequences based recommendation" ([Felfernig et al 2019:295](zotero://open-pdf/library/items/PZH6FFNA?page=11))
> "SEQREQ (Sequences based Recommendation) which recommends items based on sequential pattern mini" ([Felfernig et al 2019:295](zotero://open-pdf/library/items/PZH6FFNA?page=11))
> "CONFREQ: Recommendations for configurators" ([Felfernig et al 2019:297](zotero://open-pdf/library/items/PZH6FFNA?page=13))
> "DIAGREQ: recommending diagnoses" ([Felfernig et al 2019:300](zotero://open-pdf/library/items/PZH6FFNA?page=16))
pages/ReadingNotes/PRIN2021/khomhSoftwareEngineeringMachineLearning2018 - Gray Annotations (18122020, 001417).md
+6
View File
@@ -0,0 +1,6 @@
* Mdnotes File Name: [[khomhSoftwareEngineeringMachineLearning2018]]
# Gray Annotations (18/12/2020, 00:14:17)
> "difficulty of testing ML and AI systems." ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
pages/ReadingNotes/PRIN2021/khomhSoftwareEngineeringMachineLearning2018 - Green Annotations (18122020, 001417).md
+56
View File
@@ -0,0 +1,56 @@
* Mdnotes File Name: [[khomhSoftwareEngineeringMachineLearning2018]]
# Green Annotations (18/12/2020, 00:14:17)
> "we still experience failures and shortcomings in the resulting soft­ ware systems. The main reason is the shift in the development paradigm in­ duced by ML and AI." ([Khomh et al 2018:81](zotero://open-pdf/library/items/5486JT7B?page=1))
> "with ML techniques, these rules are inferred from training data (from which the requirements are gener­ ated inductively)." ([Khomh et al 2018:81](zotero://open-pdf/library/items/5486JT7B?page=1))
> "This paradigm shift makes reasoning about the be­ havior of software systems with ML components difficult, resulting in software systems that are intrinsi­ cally challenging to test and verify." ([Khomh et al 2018:81](zotero://open-pdf/library/items/5486JT7B?page=1))
> "the learned behavior of an ML­based system might be incorrect, even if the learning algorithm is imple­ mented correctly, a situation in which traditional testing techniques are ineffective." ([Khomh et al 2018:81](zotero://open-pdf/library/items/5486JT7B?page=1))
> "critical problem is how to effectively develop, test, and evolve such systems, given that they don't have (complete) specifications or even source code corresponding to some of their critical behaviors." ([Khomh et al 2018:81](zotero://open-pdf/library/items/5486JT7B?page=1))
*TESTING (ML) SYSTEMS THAT LACK SPECIFICATIONS OR EVEN SOURCE CODE ([note on p.81](zotero://open-pdf/library/items/5486JT7B?page=1))*
 
> "AI technology's strength comes from the ability to abstract up from different factors of varia­ tion between environments, to obtain models that can general­ ize and transfer to situations that weren't encountered before" ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
> "AI tech­ nologies' main challenge is" ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
> "he need for sufficient, labeled data to cover all important factors (features) of a given problem." ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
> "AI, in fact, needs more training data than humans do!" ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
> "appli­ cations of AI still risk being limited to domains in which labeled data is cheap." ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
> "instead of touting a "100 percent self­driving car," auto­ motive companies should advertise their products as "AI­assisted cars," with a clear list of the ways in which AI is assisting." ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
> "If a traditional computer science algorithm can solve a problem, we should just use that." ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
> "how can we perform adequate quality assurance (QA) of AI models, given that the number of environments in which the mod­ els will be deployed is unlimited and that the human operator will re­ quire a detailed explanation of any failures?" ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
> "use AI tech­ nology to reduce the search space of the environments to be tested" ([Khomh et al 2018:82](zotero://open-pdf/library/items/5486JT7B?page=2))
> "AI impacts the hu­ mans' recommendations, those rec­ ommendations are also a human filter for AI failures." ([Khomh et al 2018:83](zotero://open-pdf/library/items/5486JT7B?page=3))
> "Creating an efficient syntax for automatic differentiation that can deliver ease of implementation, per­ formance, usability, and flexibility is important but difficult." ([Khomh et al 2018:83](zotero://open-pdf/library/items/5486JT7B?page=3))
*CHALLENGES ([note on p.83](zotero://open-pdf/library/items/5486JT7B?page=3))*
 
> "esting and debugging these implementations are also salient challenges." ([Khomh et al 2018:83](zotero://open-pdf/library/items/5486JT7B?page=3))
> "How should software develop­ ment teams integrate the AI model lifecycle (training, testing, deploying, evolving, and so on) into their software process?" ([Khomh et al 2018:84](zotero://open-pdf/library/items/5486JT7B?page=4))
> "What new roles, artifacts, and activities come into play, and how do they tie into existing agile or DevOps processes?" ([Khomh et al 2018:84](zotero://open-pdf/library/items/5486JT7B?page=4))
> "testing" ([Khomh et al 2018:84](zotero://open-pdf/library/items/5486JT7B?page=4))
> "intersections" ([Khomh et al 2018:84](zotero://open-pdf/library/items/5486JT7B?page=4))
> "critical challenges of as­ suring the quality of AI and software systems in general." ([Khomh et al 2018:84](zotero://open-pdf/library/items/5486JT7B?page=4))
pages/ReadingNotes/PRIN2021/sculleyHiddenTechnicalDebt - Green Annotations (18122020, 184601).md
+82
View File
@@ -0,0 +1,82 @@
* Mdnotes File Name: [[sculleyHiddenTechnicalDebt]]
# Green Annotations (18/12/2020, 18:46:01)
> "Undeclared Consumers. Oftentimes, a prediction from a machine learning model ma is made widely accessible, either at runtime or by writing to files or logs that may later be consumed by other systems. Without access controls, some of these consumers may be undeclared, silently using the output of a given model as an input to another system. In more classical software engineering, these issues are referred to as visibility debt [13]." ([Sculley et al :10](zotero://open-pdf/library/items/TJI3FLRY?page=2))
> "It may be surprising to the academic community to know that only a tiny fraction of the code in many ML systems is actually devoted to learning or prediction - see Figure 1. In the language of Lin and Ryaboy, much of the remainder may be described as "plumbing" [11]." ([Sculley et al :12](zotero://open-pdf/library/items/TJI3FLRY?page=4))
> "Using generic packages often results in a glue code system design pattern, in which a massive amount of supporting code is written to get data into and out of general-purpose packages." ([Sculley et al :13](zotero://open-pdf/library/items/TJI3FLRY?page=5))
> "Because a mature system might end up being (at most) 5% machine learning code and (at least) 95% glue code, it may be less costly to create a clean native solution rather than re-use a generic package." ([Sculley et al :13](zotero://open-pdf/library/items/TJI3FLRY?page=5))
> "An important strategy for combating glue-code is to wrap black-box packages into common API's. This allows supporting infrastructure to be more reusable and reduces the cost of changing packages." ([Sculley et al :13](zotero://open-pdf/library/items/TJI3FLRY?page=5))
> "he resulting system for preparing data in an ML-friendly format may become a jungle of scrapes, joins, and sampling steps, often with intermediate files output." ([Sculley et al :13](zotero://open-pdf/library/items/TJI3FLRY?page=5))
> "Managing these pipelines, detecting errors and recovering from failures are all difficult and costly [1]." ([Sculley et al :13](zotero://open-pdf/library/items/TJI3FLRY?page=5))
> "Abstraction Debt." ([Sculley et al :13](zotero://open-pdf/library/items/TJI3FLRY?page=5))
> "The above issues highlight the fact that there is a distinct lack of strong abstractions to support ML systems." ([Sculley et al :13](zotero://open-pdf/library/items/TJI3FLRY?page=5))
> "What is the right interface to describe a stream of data, or a model, or a prediction?" ([Sculley et al :13](zotero://open-pdf/library/items/TJI3FLRY?page=5))
*THAT'S THE KEY!!!! ([note on p.13](zotero://open-pdf/library/items/TJI3FLRY?page=5))*
 
> "distributed learning in particular, there remains a lack of widely accepted abstractions." ([Sculley et al :13](zotero://open-pdf/library/items/TJI3FLRY?page=5))
*DISTRIBUTED LEARNING ([note on p.13](zotero://open-pdf/library/items/TJI3FLRY?page=5))*
 
> "The lack of standard abstractions makes it all too easy to blur the lines between components." ([Sculley et al :14](zotero://open-pdf/library/items/TJI3FLRY?page=6))
> "using multiple languages often increases the cost of effective testing and can increase the difficulty of transferring ownership to other individuals." ([Sculley et al :14](zotero://open-pdf/library/items/TJI3FLRY?page=6))
*THAT SUPPORTS THE NEED FOR DOMAIN-SPECIFIC LANGUAGES ([note on p.14](zotero://open-pdf/library/items/TJI3FLRY?page=6))*
 
> "Another potentially surprising area where debt can accumulate is in the configuration of machine learning systems. Any large system has a wide range of configurable options, including which features are used, how data is selected, a wide variety of algorithm-specific learning settings, potential preor post-processing, verification methods, etc." ([Sculley et al :14](zotero://open-pdf/library/items/TJI3FLRY?page=6))
> "n a mature system which is being actively developed, the number of lines of configuration can far exceed the number of lines of the traditional code. Each configuration line has a potential for mistakes." ([Sculley et al :14](zotero://open-pdf/library/items/TJI3FLRY?page=6))
> "It should be easy to specify a configuration as a small change from a previous configuration. • It should be hard to make manual errors, omissions, or oversights. • It should be easy to see, visually, the difference in configuration between two models. • It should be easy to automatically assert and verify basic facts about the configuration: number of features used, transitive closure of data dependencies, etc. • It should be possible to detect unused or redundant settings. • Configurations should undergo a full code review and be checked into a repository." ([Sculley et al :14](zotero://open-pdf/library/items/TJI3FLRY?page=6))
*Support for domain-specific language!!!! ([note on p.14](zotero://open-pdf/library/items/TJI3FLRY?page=6))*
 
> "Thus if a model updates on new data, the old manually set threshold may be invalid. Manually updating many thresholds across many models is time-consuming and brittle. One mitigation strategy for this kind of problem appears in [14], in which thresholds are learned via simple evaluation on heldout validation data." ([Sculley et al :15](zotero://open-pdf/library/items/TJI3FLRY?page=7))
> "Unit testing of individual components and end-to-end tests of running systems are valuable, but in the face of a changing world such tests are not sufficient to provide evidence that a system is working as intended." ([Sculley et al :15](zotero://open-pdf/library/items/TJI3FLRY?page=7))
> "data replaces code in ML systems" ([Sculley et al :15](zotero://open-pdf/library/items/TJI3FLRY?page=7))
> "code should be tested, then it seems clear that some amount of testing of input data is critical to a well-functioning system" ([Sculley et al :15](zotero://open-pdf/library/items/TJI3FLRY?page=7))
*Adversial Machine Learning??? ([note on p.15](zotero://open-pdf/library/items/TJI3FLRY?page=7))*
 
> "non-determinism inherent in parallel learning" ([Sculley et al :16](zotero://open-pdf/library/items/TJI3FLRY?page=8))
> "Most of the use cases described in this paper have talked about the cost of maintaining a single model, but mature systems may have dozens or hundreds of models running simultaneously [14, 6]." ([Sculley et al :16](zotero://open-pdf/library/items/TJI3FLRY?page=8))
*THIS IS IMPORTANT ASPECTS TO BE MENTIONED AS MOTIVATION/CHALLENGE ([note on p.16](zotero://open-pdf/library/items/TJI3FLRY?page=8))*
 
> "How easily can an entirely new algorithmic approach be tested at full scale?" ([Sculley et al :16](zotero://open-pdf/library/items/TJI3FLRY?page=8))
> "maintainable ML" ([Sculley et al :16](zotero://open-pdf/library/items/TJI3FLRY?page=8))
> "better abstractions" ([Sculley et al :16](zotero://open-pdf/library/items/TJI3FLRY?page=8))
> "testing methodologies" ([Sculley et al :16](zotero://open-pdf/library/items/TJI3FLRY?page=8))
> "design patterns" ([Sculley et al :16](zotero://open-pdf/library/items/TJI3FLRY?page=8))
pages/ReadingNotes/PRIN2021/sunAIEnhancedOffloadingEdge2019 - Gray Annotations (17122020, 231422).md
+18
View File
@@ -0,0 +1,18 @@
* Mdnotes File Name: [[sunAIEnhancedOffloadingEdge2019]]
# Gray Annotations (17/12/2020, 23:14:22)
> "According to GE Digital, IIoT is estimated to unlock manufacturing savings and benefit 46 percent of the global economy [1]" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "Table 1 compares the existing works on AI applications in networks." ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "Bisio et al. [10] studied the role of context awareness in IIoT applications such as smart health, smart factory, and smart home scenarios." ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "Wang et al. [11] explored the trade-off between energy consumption and service latency in IIoT" ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "Li et al. [13] applied deep learning for IoT in the edge computing environment. They designed a scheduling algorithm to maximize the number of tasks in edge computing with guaranteed quality of service (QoS) requirements." ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "In [3], edge servers learn model parameters from data distributed at the edge nodes, using the gradient-descent method based on distributed learning, instead of sending data to the centralized cloud. They proposed a control algorithm for the trade-off between local update and global parameter aggregation to minimize loss function and under a given resource budget." ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "[1] GE Digital Report, "Everything You Need to Know about the Industrial Internet of Things," 2017; https://www.ge.com/ digital/blog/everything-you-needknow-about-industrial-internet-things" ([Sun et al 2019:74](zotero://open-pdf/library/items/WXTJVNM8?page=7))
pages/ReadingNotes/PRIN2021/sunAIEnhancedOffloadingEdge2019 - Green Annotations (17122020, 231422).md
+204
View File
@@ -0,0 +1,204 @@
* Mdnotes File Name: [[sunAIEnhancedOffloadingEdge2019]]
# Green Annotations (17/12/2020, 23:14:22)
> "The Industrial Internet of Things (IIoT) enables intelligent industrial operations by incorporating artificial intelligence (AI) and big data technologies." ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "An AI-enabled framework typically requires prompt and private cloud-based service to process and aggregate manufacturing data" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "integrating intelligence into edge computing is without doubt a promising development trend" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "edge intelligence brings heterogeneity to the edge servers, in terms of not only computing capability, but also service accuracy" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "this article we introduce an intelligent computing architecture with cooperative edge and cloud computing for IIoT" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "AI enhanced offloading framework is proposed for service accuracy maximization, which considers service accuracy as a new metric besides delay, and intelligently disseminates the traffic to edge servers or through an appropriate path to remote cloud." ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "performance gain of the proposed framework" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "interconnects a multitude of industrial devices, actuators, and people at work" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "IIoT incorporates artificial intelligence (AI) technologies to process and analyze data from various sources and make advanced predictive analytics, such as fault class prediction, predictive maintenance, demand forecasting." ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "smart manufacturing is a large connected and complex industrial process, which produces a large amount of multi-feature data, it is difficult to construct its operation process with an accurate mathematical model" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "AI algorithms are able to extract critical features without in-depth physical understanding of the concerned system" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "in IIoT, predictive maintenance relies on machine learning to detect anomalies in systems and then predict the failure of devices by correlating and analyzing the change in the pattern" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "IIoT typically requires prompt and private computing service to process and aggregate the manufacturing data" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "integrating intelligence into the edge is without doubt a promising development trend [2]" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "distributed computing service through small-scale data centers near the edge of the network." ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "edge computing provides real-time data analytics with privacy preserving, increases network capabilities, and avoids congestion of backbone networks and the Internet core" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "Personalization: Customized AI models can be developed at the edge servers, which are tailored to individual users' behaviors and requirements to deliver accurate results to the users." ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
*This might be related to the citizen developer / lowcode ([note on p.68](zotero://open-pdf/library/items/WXTJVNM8?page=1))*
 
> "Responsiveness: While the industrial process is time-varying and unpredictable, the computing service must be prompt and more adaptive and feasible to the new situation." ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
*This might be related to the federated learning aspect mentioned by Luciano ([note on p.68](zotero://open-pdf/library/items/WXTJVNM8?page=1))*
 
> "Privacy Preserving: Especially for IIoT, the processing information owned by industrial companies may not be willing to transmit to the remote cloud for privacy issues; thus, the edge server provides private service." ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "The AI service deployed on edge servers exhibits heterogeneity in terms of service accuracy due to the limited and heterogeneous computing capability of edge servers." ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "The impacts of edge intelligence on computing offloading remains untouched." ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "intelligent computing architecture with cooper-" ([Sun et al 2019:68](zotero://open-pdf/library/items/WXTJVNM8?page=1))
> "ative edge and cloud computing for IIoT. Then" ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "AI enhanced offloading framework for service accuracy maximization is developed" ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "service accuracy as a new metric besides delay," ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "intelligently disseminates the traffic to edge servers or through appropriate paths to remote cloud." ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "Machine learning, as an application of AI, gives devices or computer systems the ability to "learn" with data without being explicitly programmed [2]." ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "supervised learning conducts classification or regression tasks from labeled data" ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "unsupervised learning categorizes the unlabeled data into clusters" ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "Reinforcement learning indicates agents to take actions so as to maximize the cumulative reward" ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "In IIoT, machine learning algorithms are leveraged to analyze the complex manufacturing data and believer insights about predictive maintenance, industrial prognostics, and demand forecasting." ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "Generally, cloud computing is employed for data processing. However, it is difficult to transmit huge amounts of data to the remote cloud; thus, approximate and distributed computing service becomes necessary." ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
*THIS IS A RELEVANT CASE, I.E., WHEN YOU CANNOT SEND HUDGE AMOUNT OF DATA TO THE CLOUD FOR PERFORMING DATA ANALYTICS TASKS!!! ([note on p.69](zotero://open-pdf/library/items/WXTJVNM8?page=2))*
 
> "There are edge servers that are actually designed for AI-enabled computing tasks such as the NVIDIA DGX workstation" ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
> "analyzed video streams recorded on a number of surveillance cameras" ([Sun et al 2019:69](zotero://open-pdf/library/items/WXTJVNM8?page=2))
*THIS MAKES SENSE FOR THE THE SCENARIO OF LUCIANO ON SMART BUILDING. IN SUCH CASES DATA NEED TO BE ANALYSED ON THE EDGE. ([note on p.69](zotero://open-pdf/library/items/WXTJVNM8?page=2))*
 
> "The MEC application examined the video streams, classified what were normal and abnormal patterns, and then only needed to send the stream to the backbone when a potential security issue was identified." ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
*YES! VERY IMPORTANT ([note on p.70](zotero://open-pdf/library/items/WXTJVNM8?page=3))*
 
> "AI tasks laid a heavy burden on edge servers with limited resources" ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "Industrial MDs monitor the industrial parameters, and deliver the collected data to the data center for aggregation" ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "The AI-enabled IIoT service includes self-monitoring, demand forecasting, fault detection, and workforce management." ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "The decision is fed back to the IIoT devices and executed automatically." ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "he intelligent computational architecture needs to be reshaped." ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "two-layer intelligent data center, that is, edge layer and cloud layer" ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
*THAT'S THE MAIN ARCHITECTURAL INNOVATION THAT THEY PROPOSE. ([note on p.70](zotero://open-pdf/library/items/WXTJVNM8?page=3))*
 
> "Edge Layer: It accommodates lightweight intelligent computing service for IIoT," ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "Cloud Layer: It provides powerful and comprehensive computing service for IIoT at the cost of latency and communication burden." ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "The interaction between the edge layer and the cloud layer is at the cost of additional communication on the backbone network." ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
> "how to deploy the computing service between the edge layer and cloud layer, and afterward assign the computing tasks of IIoT devices according to their requirements as well as the characteristics of heterogeneous edge servers and remote cloud, needs serious consideration." ([Sun et al 2019:70](zotero://open-pdf/library/items/WXTJVNM8?page=3))
*THAT'S ANOTHER IMPORTANT *CHALLENGES* ([note on p.70](zotero://open-pdf/library/items/WXTJVNM8?page=3))*
 
> "priority" ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "accuracy" ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "delay" ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "(ui, ai, di), where ui is the degree of urgency of MD i, that is, priority (a scalar value within (0, 1)), and ai is the acceptable accuracy of MD i, and di is the acceptable delay of MD i" ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "o choose an optimal jD i, it is imperative to estimate the delay and accuracy of offloading to available edge servers, and determine the optimal offloading option according to its requirement." ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "Different from the previous research, the offloading decision depends not only on the estimated access delay but also on the accuracy the edge server can provide." ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
*THAT'S IMPORTANT, IT IS RELATED TO THE WAY TASKS ARE DISTRIBUTED AND ASSIGNED TO THE DIFFERENT EDGE SERVERS. ([note on p.71](zotero://open-pdf/library/items/WXTJVNM8?page=4))*
 
> "near-optimal offloading framework for accuracy maximization offloading with latency constraints" ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "Step 1: (Estimate the accuracy of computing task from IIoT MD i.)" ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "Step 2: (Estimate the access delay of computing tasks from MD i.)" ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "Step 3: (Offload to the appropriate edge servers.) According to the estimated accuracy and dela" ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "For those MDs that did not find an appropriate edge server or remote server, it can be accomplished locally at the CPU of MDs. For those computing tasks with predicted delay to the remote cloud lower than its delay requirement, we tend to route the computing tasks to the remote cloud, since the remote cloud is most powerful and can provide the highest accuracy. Thus, through Step 3, the accuracy of MDs is deter" ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "By the proposed offloading framework, traffic will be disseminated intelligently, according to its requirement, to the optimal edge servers or to the remote cloud through an appropriate path so that the pressure on the backbone network will be effectively alleviated." ([Sun et al 2019:71](zotero://open-pdf/library/items/WXTJVNM8?page=4))
> "y edge servers and therefore does not consume much bandwidth. Unlike large-scale training in remote cloud, transfer training requires only a small amount of targeted training data to achieve high accuracy of the network." ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
> "trAnsfEr lEArnIn" ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
> "IoT MDs turn to the edge servers for image processing to monitor the industrial process." ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
*SIMILE SCENARIO PER NOI??? ([note on p.72](zotero://open-pdf/library/items/WXTJVNM8?page=5))*
 
> "The transfer-learning-based computing and offloading framework is done following five phases, that is, training the source neural network with large-scale data in the remote cloud, loading the pretrained neural network, customizing the predictive model, training the predictive model with small-scale data in the edge, and offloading tasks to appropriate edge servers." ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
> "First, the source neural networks are trained in large-scale data in remote cloud" ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
> "Then an edge server loads the pretrained neural network from remote cloud" ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
> "he pretrained network then transforms to a customized predictive model" ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
> "Finally, we assess the service accuracy of the predictive model and offload the computing tasks of IIoT devices according to AMLC." ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
> "In order to develop lightweight machine learning technologies on edge servers in IIoT, transfer learning is adopted." ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
> "Transfer learning is a popular approach in deep learning where pre-trained models are used as the starting point to learn a new task" ([Sun et al 2019:72](zotero://open-pdf/library/items/WXTJVNM8?page=5))
*IMPORTANT FOR THE THINGS ABOUT FEDERATED VS DISTRIBUTED LEARNING MENTIONED BY LUCIANO ([note on p.72](zotero://open-pdf/library/items/WXTJVNM8?page=5))*
 
> "This shows that at the edge layer, the customized predictive model on edge servers differs from each other due to different local data, even when the pretrained networks are the same." ([Sun et al 2019:73](zotero://open-pdf/library/items/WXTJVNM8?page=6))
> "we find it feasible to deploy machine learning applications to edge servers and employ service accuracy as a metric in the traffic offloading of MEC, while it also faces many challenges, such as the storage of training and test data, model training, and parameter updates" ([Sun et al 2019:73](zotero://open-pdf/library/items/WXTJVNM8?page=6))
*IMPORTANT CHALLENGES!!! ([note on p.73](zotero://open-pdf/library/items/WXTJVNM8?page=6))*
 
> "in AI-enabled edge computing, it is also challenging to appropriately tailor the AI-based computing service to trade off between accuracy and the constrained computing resources." ([Sun et al 2019:73](zotero://open-pdf/library/items/WXTJVNM8?page=6))
*CHALLENGE ([note on p.73](zotero://open-pdf/library/items/WXTJVNM8?page=6))*
 
> "proposed an intelligent computing architecture in IIoT with cooperation between edge servers and remote cloud." ([Sun et al 2019:74](zotero://open-pdf/library/items/WXTJVNM8?page=7))
*THAT'S THE IDEA OF THE PROPOSED APPROACH ([note on p.74](zotero://open-pdf/library/items/WXTJVNM8?page=7))*
 
> "AI-driven offloading framework considering service accuracy as a new metric, intelligently disseminating traffic to edge servers or remote cloud" ([Sun et al 2019:74](zotero://open-pdf/library/items/WXTJVNM8?page=7))
> "AI-based computing service to trade off between accuracy and the constrained computing resources." ([Sun et al 2019:74](zotero://open-pdf/library/items/WXTJVNM8?page=7))
*THAT'S A POSSIBLE FUTURE WORK ([note on p.74](zotero://open-pdf/library/items/WXTJVNM8?page=7))*
pages/ReadingNotes/PRIN2021/sunConvergenceRecommenderSystems2020 - Gray Annotations (18122020, 154124)According to Cisco reports, nearly 850 ZB data will be generated each year by 20.md
+14
View File
@@ -0,0 +1,14 @@
* Mdnotes File Name: [[sunConvergenceRecommenderSystems2020]]
# Gray Annotations (18/12/2020, 15:41:24)
> "According to Cisco reports, nearly 850 ZB data will be generated each year by 2021, but the data center is only 20.6 ZB [10]." ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
QUESTO E' FONDAMENTALE PER MOTIVARE LA QUESTIONE EDGE E FEDERATED LEARNING ([note on p.47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "It indicates that the location distribution of data sources is undergoing a transformation from data centers to an expanding number of edge devices" ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "Nowadays, various machine learning-based intelligent services have been deployed at edge servers to meet the critical requirements (e.g., agility, heterogeneous data analysis, and privacy-policy strategy) of computation tasks [16][18]." ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
SUPPORTO MOTIVAZIONE ([note on p.47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
pages/ReadingNotes/PRIN2021/sunConvergenceRecommenderSystems2020 - Green Annotations (18122020, 154124)recommender systems have been used as an effective technology to lter useless i.md
+56
View File
@@ -0,0 +1,56 @@
* Mdnotes File Name: [[sunConvergenceRecommenderSystems2020]]
# Green Annotations (18/12/2020, 15:41:24)
> "recommender systems have been used as an effective technology to lter useless information and attempt to recommend the most useful items" ([Sun et al 2020:47118](zotero://open-pdf/library/items/YWLPYTI3?page=1))
USE OF RECSYS ([note on p.47118](zotero://open-pdf/library/items/YWLPYTI3?page=1))
> "Mobile edge computing is a novel computing paradigm via pushing computation/storage resource from the remote cloud servers to the network edge servers to provide more intelligent and personalized service." ([Sun et al 2020:47118](zotero://open-pdf/library/items/YWLPYTI3?page=1))
> "collaborative ltering (CF)" ([Sun et al 2020:47118](zotero://open-pdf/library/items/YWLPYTI3?page=1))
> "content-based recommendation (CB)" ([Sun et al 2020:47118](zotero://open-pdf/library/items/YWLPYTI3?page=1))
> "nowledge-based recommendation (KB)" ([Sun et al 2020:47118](zotero://open-pdf/library/items/YWLPYTI3?page=1))
> "hybrid recommendation (HR)" ([Sun et al 2020:47118](zotero://open-pdf/library/items/YWLPYTI3?page=1))
> "CF achieves the best accuracy of predictions about how much someone is going to enjoy a movie in Netix Prize, however, it has sparseness and cold-start problems [6]" ([Sun et al 2020:47118](zotero://open-pdf/library/items/YWLPYTI3?page=1))
> "data sources contain the following key characteristics." ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "Sparsity: in the edge environment, the historical data sources stored in edge server comes from a small amount or even one user's proles" ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "Heterogeneity: edge devices are produced by companies around the world" ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "Mobility: the mobility is an inherent characteristic of users in the mobile networks" ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "Volatility: the state of the mobile edge network is volatility, when one user invokes a service many times, the QoS data may be different each time." ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "recommender systems based on cloud computing have been proposed in traditional Internet environments, they are gradually unable to deal with these novel emerging services and massively distributed data in mobile edge network, they may fail to predict what users' interests and demands are." ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
THIS IS A MOTIVATION STATEMENT!
CHALLENGE ([note on p.47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "1) cold-start problem, as data sources of active users are usually very sparse, even new or inactive users lack relevant proles, the cold-start problem occurs" ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "2) exploration and exploitation problem, for example, in online shopping, exploration implies recommending new goods and exploitation entails reusing existing goods. How to nd an optimal trade-off between exploration and exploitation is crucial issu" ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> ") security and privacy problem, the data sources are produced by various IoT devices and distributed at different edge platforms, resulting in potential leakage of user data security problem" ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "Mobile edge computing (MEC)" ([Sun et al 2020:47119](zotero://open-pdf/library/items/YWLPYTI3?page=2))
> "four enabling technologies for building recommender systems and Edge computing:" ([Sun et al 2020:47120](zotero://open-pdf/library/items/YWLPYTI3?page=3))
> ") Recommender systems on Edge" ([Sun et al 2020:47120](zotero://open-pdf/library/items/YWLPYTI3?page=3))
> "2) Recommender systems in Edge" ([Sun et al 2020:47120](zotero://open-pdf/library/items/YWLPYTI3?page=3))
> "3) Edge computing for recommender systems" ([Sun et al 2020:47120](zotero://open-pdf/library/items/YWLPYTI3?page=3))
> "4) Recommender systems for Edge computing" ([Sun et al 2020:47120](zotero://open-pdf/library/items/YWLPYTI3?page=3))
> "However, conventional recommender systems are gradually unable to meet the requirements of IoT services. Recently, a novel computing paradigm has been proposed by pushing computation and storage resources from the central cloud servers to network edges. Hence, deploying recommender systems applications at the edge servers can perform some lightweight processing to improve QoS." ([Sun et al 2020:47125](zotero://open-pdf/library/items/YWLPYTI3?page=8))
pages/ReadingNotes/PRIN2021/trakadasArtificialIntelligenceBasedCollaboration2020 - Gray Annotations (18122020, 150812).md
+44
View File
@@ -0,0 +1,44 @@
* Mdnotes File Name: [[trakadasArtificialIntelligenceBasedCollaboration2020]]
# Gray Annotations (18/12/2020, 15:08:12)
> "Industry 4.0 concepts are expected to significantly increase their footprint in industrial sectors by 20% in the next five years, since they allow leaner and more ecient production [4,5]" ([Trakadas et al 2020:5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))
> "In this context, many manufacturing companies are interested in accelerating the adoption and integration of secure, trustworthy artificial intelligence (AI) [6]" ([Trakadas et al 2020:5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))
> "AI-based manufacturing has the potential to improve the business key performance indicators (KPIs) of manufacturing processes by leveraging heterogeneous industrial big data analysis, information modelling and federation [7-9]" ([Trakadas et al 2020:5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))
> "interconnection of AI-based manufacturing processes with currently deployed wireless networks is a challenging research field, especially when central processing is performed outside industrial premises [10,11]" ([Trakadas et al 2020:5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))
> "AI is critical to the cybersecurity aspect of an IIoT-enabled connected manufacturing environment, for accurately detecting and mitigating threats [16-19]" ([Trakadas et al 2020:5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))
> "Industry 4.0 will make machines increasingly smarter by using AI models [28]" ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
*REFERENCE IMPORTANTE CHE MOTIVA ASPETTI DI INTEROPERABILITA' ([note on p.5483](zotero://open-pdf/library/items/K7JAGWW6?page=4))*
 
> "TensorFlow Federated provide support for decentralized AI models learning or computation over locally controlled data sources [39]." ([Trakadas et al 2020:5486](zotero://open-pdf/library/items/K7JAGWW6?page=7))
> "adopting this novel approach [40], this scheme heavily reduces the administrative eort for key sharing and management, while ensuring end-to-end information protection" ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
> "adversarial attacks, which is the core element of trustworthy AI, has recently received much attention [41]." ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
> "The creation of models of the state and behavior can be done in a semi-automatic manner, using a newly devised AutoML tool that takes as input vector representations of sequential input data [46]" ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
*AUTOML ([note on p.5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))*
 
> "Currently, federated learning is being adopted in dierent scenarios such as banking [61] and healthcare [62]." ([Trakadas et al 2020:5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))
*HERE WE HAVE COUPLE OF CONCRETE SCNEARIOS USING FEDERATED LEARNING ([note on p.5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))*
 
> "There are currently many research activities to develop faster or more resource ecient PSI protocols [38,64]." ([Trakadas et al 2020:5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))
> "26. Sittón-Candanedo, I.; Alonso, R.C.; Rodríguez-González, S.; Alberto García Coria, J.; De La Prieta, F. Edge Computing Architectures in Industry 4.0: A General Survey and Comparison. In International Workshop on Soft Computing Models in Industrial and Environmental Applications; Springer: Cham, Switzerland, 2019; Volume 950." ([Trakadas et al 2020:5497](zotero://open-pdf/library/items/K7JAGWW6?page=18))
> "Comiter, M. Attacking Artificial Intelligence, AI's Security Vulnerability and What Policymakers about It. In Belfer Center Paper; Harvard Kenedy School: Cambridge, MA, USA, 2019." ([Trakadas et al 2020:5498](zotero://open-pdf/library/items/K7JAGWW6?page=19))
pages/ReadingNotes/PRIN2021/trakadasArtificialIntelligenceBasedCollaboration2020 - Green Annotations (18122020, 150812).md
+318
View File
@@ -0,0 +1,318 @@
* Mdnotes File Name: [[trakadasArtificialIntelligenceBasedCollaboration2020]]
# Green Annotations (18/12/2020, 15:08:12)
> "important enablers for AI adoption in manufacturing contexts and is promoted in terms of business intelligence optimization, human-in-the-loop and secure federation across manufacturing sites" ([Trakadas et al 2020:5480](zotero://open-pdf/library/items/K7JAGWW6?page=1))
> "(1) components that extend the functionality of the existing layers in the Reference Architectural Model for Industry 4.0; (2) definition of new layers for collaboration by means of human-in-the-loop and federation; (3) security concerns with AI-powered mechanisms." ([Trakadas et al 2020:5480](zotero://open-pdf/library/items/K7JAGWW6?page=1))
> "most AI techniques are based on mathematical models that are dicult to understand by the general public, so most people use AI-based technology as a black box that they eventually start to trust based on their personal experience" ([Trakadas et al 2020:5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))
> "The application of human-centric AI (HAI) in internet of things (IoT) systems, so that IoT systems cannot only learn from users but also provide easy-to-understand explanations about decisions or estimations is a new research field [12]." ([Trakadas et al 2020:5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))
*ASPETTI MOTIVAZIONALI MOLTO IMPORTANTI. ([note on p.5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))*
 
> "At the same time, introducing AI will lead to a more productive and safer working space, relieving human workers from routine procedures and employing intelligent machines and robots to perform heavy tasks, thus allowing human workers to focus on creativity, reasoning and decision making [20,21]." ([Trakadas et al 2020:5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))
*MOTIVATION ([note on p.5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))*
 
> "to provide an IIoT-based system that increases performance and safety in the manufacturing domain." ([Trakadas et al 2020:5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))
*MAIN GOAL ([note on p.5481](zotero://open-pdf/library/items/K7JAGWW6?page=2))*
 
> "AI solutions are implemented in dispersed and isolated components of manufacturing IT systems." ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
> "Current Industry 4.0 reference architectures do not properly integrate the needed building blocks such as new deployment paradigms (e.g., edge-based learning to reduce bandwidth load on the enterprise network)" ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
> "scalable data-processing pipelines and information models" ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
> "AI-enabled digital twins used for monitoring and optimizing business intelligence [24-26]" ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
*IMPORTANT MOTIVATIONS AND CONTEXT DESCRIPTION ([note on p.5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))*
 
> "availability of big data has been one of the most important enablers for the recent wave of AI innovations [27]" ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
> "Moreover, every phase of AI algorithm design requires high-level skills (model selection, training, hyperparameter optimization). In the agile manufacturing of the future, these costs must be amortized over low-volume batches (even lot-size-one)." ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
*IMPORTANT MOTIVATIONS AND CONTEXT DESCRIPTION ([note on p.5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))*
 
> "AI technologies should not only be used for data analytics in support of business intelligence, but also for automated decision making on manufacturing process parameters and configurations." ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
> "AI algorithms are often black-box models (e.g., deep learning), while the inner workings of an algorithm fetched from a remote repository are not fully understood or the decisions of one algorithm create a conflict with other algorithmic decisions" ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
> "Proper secure federation mechanisms and AI-based cyberattack risk analysis are crucial cross-cutting concerns in AI-based manufacturing systems." ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
> "the performance of modern AI techniques requires large volumes of high-quality data which are often not available inside a single enterpris" ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
*MOTIVA BENE LA QUESTIONE DI LUCIANO, PERCHE' SI HA BISOGNO DI FEDERATED LEARNING ([note on p.5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))*
 
> "AI techniques will be used to extend and improve the levels of communication and collaboration between computer systems and human workers." ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
> "new intelligent design and decision-making tools must be developed to promote human agency and oversight, simplifying the understanding and usage of AI results and considering multiple collaboration schemes depending on the situation. Human-AI will work in tandem in any phase of the product construction process, from design, over intelligent manufacturing execution monitoring to predictive maintenance." ([Trakadas et al 2020:5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))
*OBIETTIVI ([note on p.5482](zotero://open-pdf/library/items/K7JAGWW6?page=3))*
 
> "collaborative and intelligent factory of the future." ([Trakadas et al 2020:5484](zotero://open-pdf/library/items/K7JAGWW6?page=5))
> "ration with other manufacturing sites, while security is aboration and for cross-cutting concern. In the following, we describe the functionality of each component as shown inis a cross-cutting Figure 2 concern" ([Trakadas et al 2020:5484](zotero://open-pdf/library/items/K7JAGWW6?page=5))
*LUCIANO ([note on p.5484](zotero://open-pdf/library/items/K7JAGWW6?page=5))*
 
> "the factory-wide datasets so that the components of the upperyersistoperform layers can make decisions based on the outcome of AI algorithms running on top of such processedrlayerscanmake data streams or batch datasets. Raw (non-labeled) data generated by manufacturing devices ared data streams or processed alo batchdatasets" ([Trakadas et al 2020:5484](zotero://open-pdf/library/items/K7JAGWW6?page=5))
> "re functionality to deploy AI algorithms closer to the sensor (edgeer to the sensor computing) and to detect shifts in the dataset statistics, indicating a need to retrain algorithms. Inetrainalgorithms" ([Trakadas et al 2020:5484](zotero://open-pdf/library/items/K7JAGWW6?page=5))
> "AI-enabled data pipelines orchestrator component enables the creation and deployment ofnddeploymentof data processing pipelines, with two major objectivemajorobjectives:" ([Trakadas et al 2020:5484](zotero://open-pdf/library/items/K7JAGWW6?page=5))
> "(i) the component should allow the possibilitythepossibilityto to set up pipelines consisting of typical data processing tasks (feature conversion, feature reduction,rereduction,data data anonymization and fusion, data cleaning, labeling and annotation, etc.) and AI models (used inodels(usedin the the services of the upper latheupperlayers)" ([Trakadas et al 2020:5484](zotero://open-pdf/library/items/K7JAGWW6?page=5))
> "e factory; (ii) thei)thecomponent component should allow the deployment of the pipelines and the orchestration of the differentcomponents" ([Trakadas et al 2020:5484](zotero://open-pdf/library/items/K7JAGWW6?page=5))
> "automate the deployment process on distributed infrastructure (edge device, edge cloud, public cloud) and to orchestrate the dierent modules and exact frameworks needed to run the processes." ([Trakadas et al 2020:5485](zotero://open-pdf/library/items/K7JAGWW6?page=6))
*IMPORTANTE: RELATIVO AL AI PIPELINES (DESIGN AND DEPLOYMENT) ([note on p.5485](zotero://open-pdf/library/items/K7JAGWW6?page=6))*
 
> "Edge-based learning is required for latency-sensitive situations and/or when upstream bandwidth is insucient, e.g., audio and video from an augmented reality (AR) headset or processing light detection and ranging (LIDAR) data on a mobile robot. The component will support novel neural network architectures that can be trained without requiring large amounts of labelled data and that are resource-ecient [36]." ([Trakadas et al 2020:5485](zotero://open-pdf/library/items/K7JAGWW6?page=6))
*IMPORTANTE LUCIANO ([note on p.5485](zotero://open-pdf/library/items/K7JAGWW6?page=6))*
 
> "threat intelligence manager takes advantage of the collected and curated datasets and applies AI algorithms for executing threat analysis in order not only to predict potential cybersecurity incidents but most importantly to manage and mitigate such incidents in a timely manner." ([Trakadas et al 2020:5485](zotero://open-pdf/library/items/K7JAGWW6?page=6))
> "4.2. Functional and Business Intelligence" ([Trakadas et al 2020:5485](zotero://open-pdf/library/items/K7JAGWW6?page=6))
> "mirror the state of the production process in digital twins, including the logic that determines the transition to other production steps or states." ([Trakadas et al 2020:5485](zotero://open-pdf/library/items/K7JAGWW6?page=6))
> "This layer provides innovative tools that will facilitate intuitive and ecient collaboration between humans, machines and AI systems allowing them to take advantage of each other 's strengths for more eective cooperative and intuitive task execution and decision making" ([Trakadas et al 2020:5486](zotero://open-pdf/library/items/K7JAGWW6?page=7))
> "Federated Learning component aims at solving the problem of data collection for feeding or training AI models, while assuring the ownership and confidentiality of the data. In manufacturing, most (if not all) data and information are confidential because they relate directly to details of the production process, product characteristics, volumes, etc." ([Trakadas et al 2020:5486](zotero://open-pdf/library/items/K7JAGWW6?page=7))
> "The inter-manufacturing knowledge exchange serves as an interface for knowledge exchange across manufacturing sites or distinct manufacturing processes." ([Trakadas et al 2020:5486](zotero://open-pdf/library/items/K7JAGWW6?page=7))
> "ince there is a need to control what information is exposed and exchanged, rather than allowing open access to the local knowledge repository, this component contains a query engine to handle external requests. Such query engines also enable the realization of a federated query-processing mechanism over multiple sites." ([Trakadas et al 2020:5486](zotero://open-pdf/library/items/K7JAGWW6?page=7))
*FEDERATED QUERY-RPOCESSING METCHANISM OVER MULTIPLE SITES ([note on p.5486](zotero://open-pdf/library/items/K7JAGWW6?page=7))*
 
> "4.5. Security and Authorization" ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
> "security and authorization requirements on information and data sharing." ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
> "Cybersecurity for Artificial Intelligence (AI)" ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
> "Artificial intelligence attacks, i.e., attacks on the AI algorithm, can take two forms: input attacks and poisoning attacks. The former consists in manipulating the input to the AI system during the operation phase so that it delivers the wrong results. Input attacks are relatively easy to launch and succeed since they do not require a manipulated AI system. Poisoning attacks, on the other hand, have to do with the corruption of the process used to build the AI model. In this case, inaccurate or mislabeled data are provided to the model during the training phase to manipulate the learning process. This type of attack can also be launched against federated learning; in this case, manipulated data or an algorithm of a member of the federation can result in the corruption of the global model." ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
*IMPORTANT! RELATED TO AML WORK!!!
IT APPLIES ALSO IN THE CASE OF FEDERATED LEARNING ([note on p.5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))*
 
> "challenges regarding the implementation of the innovative AI-based components of the proposed system" ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
> "AI-Driven Modelling of Manufacturing Assets" ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
> "Digital twins are virtual, high-fidelity models of the current state and internal behavior of physical assets on the shop floor [25]." ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
> "there is a lack of information models and process libraries that allow users to replicate and scale their digital twins." ([Trakadas et al 2020:5487](zotero://open-pdf/library/items/K7JAGWW6?page=8))
> "AutoML" ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
*DA VEDERE AUTOML ([note on p.5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))*
 
> "For the IT resource aspect, processing all raw data on a public cloud infrastructure is an unscalable solution for many manufacturing companies, either because there is too much sensor data to upload, the latency to the cloud is prohibitive or because the sensor data is too sensitive and the company does not want to expose this. Therefore, edge computing has been proposed and several reference architectures for edge computing in Industry 4.0 have been proposed [26" ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
*LACK OF RESOURCES FOR AI IN INDUSTRY ([note on p.5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))*
 
> "Techniques to make deep neural networks available for the latter form of edge computing start to find their way into production as more user-friendly tools become available. For instance, TensorFlow Lite allows converting a trained model for deployment on microcontrollers or embedded Graphics Processor Units (GPUs)." ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
*TensorFlow Lite: Come diceva Luciano ([note on p.5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))*
 
> "5.2. Multi-Channel, Context-Aware Interaction on the Shop Floor" ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
*Relativo alla parte di NLI di Mariani ([note on p.5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))*
 
> "The latest developments [48,49] allow humans to convey information with AI systems through multiple channels by integrating advanced human-machine interfaces (gestures, facial expressions). These interfaces, which can be obtained by using 2-D and/or 3-D cameras and other sensors, such as gyroscopes or accelerometers, oer information related to the context and the situation that is relevant to the interaction." ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
*Gesture, facial expressions interaction etc ([note on p.5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))*
 
> "voice understanding" ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
> "experts in simulations and modelling are not available in SMEs and large enterprises, and that data-driven digital twins are usually trained on data collected at system level. These limitations make it economically costly to develop novel digital twins, which is often needed in modern manufacturing with agile reconfigurations." ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
*IMPORTANT - LACK OF EXPERTS ([note on p.5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))*
 
> "Data-driven digital twins are built using machine learning." ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
> "DL comes with its own problems: it requires domain knowledge to select the appropriate machine learning pipeline and it is very resource hungry in terms of computing and storage resources [36,43,44]" ([Trakadas et al 2020:5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))
*PROBLEMS ON DEEP LEARNING ([note on p.5488](zotero://open-pdf/library/items/K7JAGWW6?page=9))*
 
> "Intelligent Decision Support" ([Trakadas et al 2020:5489](zotero://open-pdf/library/items/K7JAGWW6?page=10))
> "increase the eciency in the decision-making process" ([Trakadas et al 2020:5489](zotero://open-pdf/library/items/K7JAGWW6?page=10))
> "urrent approaches fail to learn from decisions taken by humans." ([Trakadas et al 2020:5489](zotero://open-pdf/library/items/K7JAGWW6?page=10))
> "ontinuous learning functionalities" ([Trakadas et al 2020:5489](zotero://open-pdf/library/items/K7JAGWW6?page=10))
> "Threat Intelligence Manager (TIM)" ([Trakadas et al 2020:5489](zotero://open-pdf/library/items/K7JAGWW6?page=10))
> "In the case of malicious activity, various AI algorithms, such as naïve Bayes, random forests and support vector machine (SVM) have been proposed [55]." ([Trakadas et al 2020:5489](zotero://open-pdf/library/items/K7JAGWW6?page=10))
> "By exploiting advanced AI methods, the threat intelligence manager (TIM) component intends to model the dynamic interactions of Industry 4.0 subsystems and discover known and unknown attacks, while surpassing existing signatureand anomaly-based methods." ([Trakadas et al 2020:5489](zotero://open-pdf/library/items/K7JAGWW6?page=10))
> "Federated AI across Manufacturing Sites" ([Trakadas et al 2020:5489](zotero://open-pdf/library/items/K7JAGWW6?page=10))
> "In this context, federated query processing [59] is an active research field dealing with techniques for proper delegation of the execution of parts of queries to specific sources" ([Trakadas et al 2020:5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))
*IMPORTANT! FEDERATED QUERY PROCESSING ([note on p.5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))*
 
> "To improve the scalability of such federations, aggregation techniques could be used, where one or more independent aggregators would continuously crawl sources, and maintain data summaries [60]." ([Trakadas et al 2020:5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))
> "Joining datasets held by dierent actors can address this issue. Oft" ([Trakadas et al 2020:5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))
*THAT'S THE GOAL TO ADDRESS THE ISSUE THAT SINGLE PARTY DOES NOT HAVE SUFFICIENTLY LARGE DATA SETS FOR TRAINING ([note on p.5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))*
 
> "federated learning provides a solution enabling machine learning over distributed and decentralized datasets." ([Trakadas et al 2020:5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))
> "federated learning opens up new business models (AI as a Service, AIaaS) to analyze data provided by a customer" ([Trakadas et al 2020:5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))
> "federated learning represents a solution to this problem enabling both the service provider and its customer to achieve their objectives while preserving the business assets." ([Trakadas et al 2020:5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))
> "t is possible to generate in an automatic manner intelligent services in parts of the building blocks or in the process as a whole." ([Trakadas et al 2020:5490](zotero://open-pdf/library/items/K7JAGWW6?page=11))
> "handling and labeling of data of very dierent types." ([Trakadas et al 2020:5491](zotero://open-pdf/library/items/K7JAGWW6?page=12))
> "dierent system components can be exploited for optimizing manufacturing logistics processes, and facilitating zero-defect manufacturing" ([Trakadas et al 2020:5491](zotero://open-pdf/library/items/K7JAGWW6?page=12))
> "The current AMR production logistics is supposed to be carried out by the system illustrated in Figure 3a" ([Trakadas et al 2020:5491](zotero://open-pdf/library/items/K7JAGWW6?page=12))
> "t is assumed that there are two issues that need to be addressed in an ecient manner: (a) unpredictable delivery times and downtime, and (b) vulnerability to network attacks." ([Trakadas et al 2020:5491](zotero://open-pdf/library/items/K7JAGWW6?page=12))
> "le and machines at high levels of safety and tenance services for their AMRs" ([Trakadas et al 2020:5492](zotero://open-pdf/library/items/K7JAGWW6?page=13))
> "reduced by avoiding heavy vehicles like forklifts or tuggers in fast-moving intralogistics areasorevencompromise during pe site safety." ([Trakadas et al 2020:5492](zotero://open-pdf/library/items/K7JAGWW6?page=13))
> "neously, AMR downtime is undesirable andAMRstothenetwork should be reduced as much as possible. entcybersecurityriskduetodynamicsoftwarevulnerabilitiesthataffect This scenario assumes that a company has implemented a flmunicationprotocols." ([Trakadas et al 2020:5492](zotero://open-pdf/library/items/K7JAGWW6?page=13))
> "Advances leveraging the proposed platform" ([Trakadas et al 2020:5492](zotero://open-pdf/library/items/K7JAGWW6?page=13))
> "learning from patterns of business process models and anomaly detection techniques" ([Trakadas et al 2020:5492](zotero://open-pdf/library/items/K7JAGWW6?page=13))
> "prediction of AMR downtime can be done using multivariate statistical models on data of key AMR system parameters" ([Trakadas et al 2020:5492](zotero://open-pdf/library/items/K7JAGWW6?page=13))
> "Edge-based (i.e., on-robot) learning can be used to reduce the amount of data uploaded via the customer network to AMR supplier 's cloud back-end and assure the confidentiality of production activity related data" ([Trakadas et al 2020:5492](zotero://open-pdf/library/items/K7JAGWW6?page=13))
> "run AI observers that collect data about typical movement patterns on the floor (e.g., edge-based learning on surveillance camera data) and about how fleet behaviors relate to production goals (e.g., from Enterprise Resource Planning—ERP and Manufacturing Execution System—MES systems)" ([Trakadas et al 2020:5492](zotero://open-pdf/library/items/K7JAGWW6?page=13))
> "As unauthorized access to machines and data might compromise the physical integrity of human workers, communication between AMR, the platform and AMR supplier 's cloud back-end can be secured by signcryption schemes provided by the system." ([Trakadas et al 2020:5493](zotero://open-pdf/library/items/K7JAGWW6?page=14))
> "The number and the diversity of characteristics of the orders do not allow for common process standardization." ([Trakadas et al 2020:5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))
*DIVERSITY ([note on p.5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))*
 
> "The data and information collected across the departments through manual and automatic procedures are solely processed by each department, without considering factory-wide optimization metrics." ([Trakadas et al 2020:5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))
*FACTORY-WIDE OPTIMIZATION METRICS ([note on p.5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))*
 
> "The deficiencies are observed, in the majority of the cases, during the quality control of the final product, thus leaving no space for corrective actions." ([Trakadas et al 2020:5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))
*NEEDS FOR EARLY CORRECTIVE ACTIONS ([note on p.5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))*
 
> "letting personnel only make the final decisions." ([Trakadas et al 2020:5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))
*LACK OF PERSONNEL KNOWLEDGE AND TRAINING ON THE USE OF INNOVATIVE TOOLS ([note on p.5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))*
 
> "The combination of data analytics and In testing and validation procedure, the goal is to define a set of KPIs per potential use case andhat can be used for validate the performance of our proposed AI-barootcause analyses" ([Trakadas et al 2020:5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))
*WHAT-IF ANALYSIS, PROJECTIONS, AND ROOT CAUSE ANALYSIS ([note on p.5494](zotero://open-pdf/library/items/K7JAGWW6?page=15))*
 
> "manufacturers will be capable of realizing agile production processes and improve the quality of products and processes" ([Trakadas et al 2020:5495](zotero://open-pdf/library/items/K7JAGWW6?page=16))
*BUSINESS IMPACTS ([note on p.5495](zotero://open-pdf/library/items/K7JAGWW6?page=16))*
 
> "more competitive in the market and thus increase their market share." ([Trakadas et al 2020:5495](zotero://open-pdf/library/items/K7JAGWW6?page=16))
> "The federated intelligence layer introduced in our approach enables new business models (e.g., high-quality AIaaS), as well as the collaboration of dierent industries towards the creation of digital twins" ([Trakadas et al 2020:5495](zotero://open-pdf/library/items/K7JAGWW6?page=16))
> "rchitecture that facilitates the collaboration between manufacturing machinery, AI and humans" ([Trakadas et al 2020:5495](zotero://open-pdf/library/items/K7JAGWW6?page=16))
*FACILITATING ARCHITECTURE!!! ([note on p.5495](zotero://open-pdf/library/items/K7JAGWW6?page=16))*
 
> "Last but not least, companies with expertise in AI for manufacturing can create significantly higher revenues by being capable of integrating their components with IoT and IT systems from dierent vendors/creators." ([Trakadas et al 2020:5495](zotero://open-pdf/library/items/K7JAGWW6?page=16))
*OPEN COLLABORATION AMONG COMPONENTS FROM DIFFERENT VENDORS/CREATIOS ([note on p.5495](zotero://open-pdf/library/items/K7JAGWW6?page=16))*
 
> "components for timely data collection" ([Trakadas et al 2020:5496](zotero://open-pdf/library/items/K7JAGWW6?page=17))
> "processing and curation" ([Trakadas et al 2020:5496](zotero://open-pdf/library/items/K7JAGWW6?page=17))
> "elying on the dynamic instantiation of data pipelines" ([Trakadas et al 2020:5496](zotero://open-pdf/library/items/K7JAGWW6?page=17))
> "while addressing security, privacy and confidentiality concerns" ([Trakadas et al 2020:5496](zotero://open-pdf/library/items/K7JAGWW6?page=17))
> "cross the physical and virtual entities" ([Trakadas et al 2020:5496](zotero://open-pdf/library/items/K7JAGWW6?page=17))
> "The collected data and information models are transformed into AI-enabled functional intelligence, leading to business knowledge, actionable insights and informed decisions, while being capable of recognizing complex events and process deviations that cannot be captured easily and in a timely way through human judgment" ([Trakadas et al 2020:5496](zotero://open-pdf/library/items/K7JAGWW6?page=17))
*MAIN GOALS/BENEFITS ([note on p.5496](zotero://open-pdf/library/items/K7JAGWW6?page=17))*
 
> "25. Tao, F.; Qi, Q.; Wang, L.; Nee, A.Y.C. Digital Twins and Cyber-Physical Systems toward Smart Manufacturing and Industry 4.0. Engineering 2019, 5, 653-661. [CrossRef]" ([Trakadas et al 2020:5497](zotero://open-pdf/library/items/K7JAGWW6?page=18))