# Geometric Deep Lean Learning: Deep Learning in Industry 4.0 Cyber–Physical Complex Networks

^{1}

^{2}

^{3}

^{4}

^{5}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Background

- Industry 4.0. The term Industry 4.0 has gained large traction since it was first publicized [6], stating the need for a paradigm shift towards a less centrally controlled manufacturing structure. It is seen as the Fourth Industrial Revolution, with the first three being mechanization through steam power, mass production through electrically operated engineering, and the digital revolution through the integration of electronics and IT. Industry 4.0 enables more production autonomy as technology becomes more interconnected, and machines are able to influence each other by creating a cyber–physical system.
- Cyber–Physical Systems. The term “cyber–physical system” in the context of Industry 4.0 refers to the tight conjoining of and co-ordination between computational and physical resources. The impact on the development of such systems is a new paradigm of technical systems based on collaborative embedded software systems [7].
- Lean Management. Lean-management systems in an Industry 4.0 cyber–physical context have been described as sociotechnical entities that aim to systematically reduce the variability of value-creation processes [8,9,10,11,12,13]. These two fundamental dimensions, the social and the technical, are subsequently meant to symbiotically support each other to maximize value creation through the systematic elimination of activities that do not add value for the client. A series of models were presented by scholars that allow the analysis and quantification of these systems as complex networks [14,15].
- Complex-Networked Organizational Design. Under the organizational-network paradigm, modern Industry 4.0 cyber–physical lean-management-oriented organizations can be understood as a symbiotic sociotechnical ecosystem of social networks [16] that interacts with increasingly complex-networked physically distributed interconnected sensors [17], whose readings are modeled as time-dependent signals on the vertices, human or cyber–physical, respectively. This means that, on the nodes of the network, attributes can be found that describe them as having the form of a given time series.Within this framework, a complex network is defined as a graph with nontrivial topological features that do not occur in simple graphs such as lattices and random networks [18]. For any given time t, lean complex cyber–physical networks can be formally described by time-dependent graphs $\Omega \left(t\right)=\left[N\right(t);E(t\left)\right]$ that can be understood as lists of $N\left(t\right)$ nodes and $E\left(t\right)\subset \left(N\right(t\left)xN\right(t\left)\right)$ edges that represent its human and cyber–physical nodes, and its standard communication edges [19]. Given the static graph in t, $\Omega \left(t\right)$, each node and edge can be characterized by a series of typically two-dimensional signals $x=[{x}_{1},\dots ,{x}_{n}]\in \left({\mathbb{R}}^{n}x{\mathbb{R}}^{m}\right)$, where n relevant parameters of the node or axis are described as the time series of m elements. In the case of nodes, signals typically represent demographic, sociological, or competence information. In case that the nodes are human, and in the case of a cyber–physical node, relevant information on the state of the cyber–physical node expressed in time series of several key performance indicators. In the case of edges, signals typically represent information referring to the quality of measurable relationships of the individual with other stakeholders of the organization; in the case of human–human or cyber–physical-to-human edges, of the time series associated with relevant key performance indicators being reported to other stakeholders. Specifically, snapshots for the time-dependent graph can be built, that is, the time-dependent graph is considered as an ordered pair of potentially different sets. A time-dependent graph considered as a sequence of static graphs is given by Expression 1.$$\Omega =[\Omega \left({t}_{1}\right),\Omega \left({t}_{2}\right),\dots ,\Omega \left({t}_{k}\right)]$$This method is most commonly used for modeling discrete time-dependent graphs, and is suitable for the time-dependent graph with a specific time structure, especially in real-time networks such as complex-networked cyber–physical systems [20]. This modelling method is assumed here, and the time sequence of static graphs is not explicitly mentioned when referring to time-dependent graphs.

## 3. Related Work

- Offline training, and decision-support learning and predicting from a global and integrated way, for example, by extracting relevant information from an organization by means of deep-learning algorithms that analyze previously labeled text in organizational categories [66]. Alternatively, by combining deep learning with other computing methods that allow for more balanced datasets and, hence, better deep-learning performance [67].
- Digital twin and augmented reality. Creating virtual environments that, by recording, visualization, and interaction with cyber–physical assets, are capable of generating necessary tagged information in real time that is fed to deep-learning algorithms [68]. The creation of digital twins in combination with deep-learning algorithms was also proposed to enable the parallel control of cyber–physical value-creating processes [69].

- Graphwise classification. For instance, in the classification of molecules [77]. In this model, atoms represent the nodes, and chemical bonds are the edges of a graph. Research aims to extract certain features that predict certain properties of the molecule. This is relevant, for instance, to the pharmaceutical companies that are in the business of drug design. Some of these properties are toxicity and water solubility. Given a graph, researchers aim to classify a molecule graph. This is analogous to classical deep-learning-based visual image classification [78].
- Vertexwise classification. For example, in a social-network domain in which nodes are people of which we have certain demographic information, a researcher aims to predict how these people will vote in the next election. The analogy in computer vision is semantic image segmentation [79] in which the pixels of an image are labeled as belonging to a certain category.
- Graph dynamics. There are also domains that are described by fixed graphs, and others in which the graph changes with time [70]. Complex-networked cyber–physical systems belong to the second class.
- Known vs. unknown domain. In some cases, the graph can be known; in others, it is only partially known, noisy, or not known at all and needs to be learned. In these cases, the researcher aims to not only learn the graph features, but also the graph itself [80].

- Spectral filtering methods.Spectral filtering methods make use of the spectral eigendecomposition of the Laplacian graph to elegantly mathematically define convolution-like operators. The fundamental limitation of the spectral construction is that it can only be used to single and static domains. This is because filter coefficients are dependent on the eigenvector- and eigenvalue-decomposition basis of the Laplacian graph, which is highly dependent on network architecture [70]. This approach is not suitable for our needs because of the dynamic characteristics of Industry 4.0 lean-management cyber–physical complex systems and their associated complex networks.
- Local filtering methods.

## 4. Geometric Deep Lean Learning Over Industry 4.0 Lean-Management Complex-Networked Cyber–Physical Systems

- Strategic organizational design. Performing classical inference problems [76].Recently, it has been shown that this classification can be considerably improved by using information about the proximity environment [83,84]. Analyzing signals on graph vertices and edges could potentially help to learn inherent structures of the graphs, such as organizational clusters, with better accuracy than that provided by topological information alone—this is a strategic challenge to which organizational design tries to respond.
- Trust and power structures. Learning hidden organizational properties.Although deep learning has been employed in a wide variety of fields of knowledge, such as modeling social influence [85] and computer vision [86,87,88], it is important to incorporate knowledge about the domain to be treated in the model. For example, in order to build a deep-learning model for the study of a network of sensors in a cyber–physical system of industry 4.0, it might be useful, in a first approximation, to choose the edge weights of the graph as a decreasing function of the distance between nodes, as this would lead to a smooth graph signal model [89]; however, this would not be suitable for a lean structural network, because adjacency does not necessarily mean similarity [14]. For this reason, the model of the graph to be used can be superimposed on other structures, instead of being a pure unconnected abstraction. In other words, the graph that represents the complex-networked cyber–physical system in an Industry 4.0 context, can be studied from different perspectives, superimposing it to a specific sociotechnical environment that helps to better understand the statistical information that it contains. As a consequence, the integration of these priors is a fundamental challenge for the success of geometric deep lean learning. Some examples are the structures of power or trust between the different actors of an organization that are fundamental variables that influence the success of an organization, but remain elusive, since they often cannot be directly measured. Geometric deep lean learning could be applied to learn these parameters as weights between the nodes of the complex organizational network.

- Convolution on non-Euclidean complex-networked cyber–physical graph time-dependent signals.As expressed in Expression 1, for weighted time-dependent directed graph $\Omega \left(t\right)$, a series of signals $x=[x\left(1\right),\dots ,x\left(n\right)]\in \left({\mathbb{R}}^{n}x{\mathbb{R}}^{m}\right)$ expressed on its human and cyber–physical nodes, and on its standard communication edges, are considered, in which components of ${x}_{a}$ reside in or are protruding from node a.For each node, we define a proximity environment given by group ${N}_{a}=\{b:(b,a)\subset E\}$ that represents set of nodes b connected with a. This ${N}_{a}$ set is characterized by an ${\mathbb{R}}^{NxN}$ matrix S called the network-translation matrix operator that defines the manifold metric. We defined S as the graph adjacency matrix, the Laplacian of the graph, or any other normalization of it, as a linear transformation to encode the structure of a graph. Without loss of generality, the singularity problem of the adjacency matrix, which is nontrivial, was not considered in this work [91]. As shown in Figure 2, group ${N}_{a}$ represents the manifold upon which the convolution acts.The Fourier decomposition of graph $\Omega \left(t\right)$ is expressed by $\widehat{x}={U}^{-1}\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}x$, where $S=\mathcal{U}\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}\Lambda \phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}{\mathcal{U}}^{-1}$ and autovalues $\Lambda $ describe the frequencies of the graph [92]. Now, we can directly filter x from the spectral domain by means of function $f:\mathbb{C}\to \mathbb{R}$ that allows to compute convolution $\widehat{z}=f(\Lambda )\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}\widehat{x}$ by means of point-by-point multiplication in the spectral domain between filter $f(\Lambda )$ and the Fourier transform of the graph in x. Therefore, by inverting the Fourier transform of the graph, we obtain the extension of the convolutional operation to the non-Euclidean time-dependent graph in Equation (2).$$\begin{array}{c}\hfill z=\mathcal{P}\left(S\right)x\text{}\mathrm{and}\text{}\mathcal{P}\left(S\right)=\mathcal{U}f(\Lambda ){\mathcal{U}}^{-1}\end{array}$$The filter operation can be directly described on the node, resulting in an alternative formulation given by Equation (3), where scalar parameter ${\varphi}_{a,b}$ is a representation of the information weights coming from neighbour node b into or from node a.$$\begin{array}{c}\hfill {z}_{a}=\sum _{b\in {N}_{a}\cup a}^{}{\varphi}_{a,b}\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}{x}_{b}\end{array}$$Due to the local properties of S, ${z}_{a}$ can be obtained in the domain of the node through local-information exchange. This means that the initial signal on the node is recursively transformed by S a K number of times until decomposition is obtained that determines ${z}_{a}$ as the convolution between the network filter with a polynomial transfer function and ${x}_{b}$.By means of the Fourier transform of the network, the screening operation of Equation (3) has the transfer function given by Equation (4):$$\begin{array}{c}\hfill h(\Lambda )=\sum _{k=0}^{\kappa}{\varphi}_{k}\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}{\Lambda}^{k}\end{array}$$This filter, based on local-information exchanges, captures information in K-radius proximity from the node representing the depth of the geometric-deep-lean-learning algorithm.Taking into account this convolutional operation given by Equation (3), we are able to compute the fth level feature produced as output of the lth layer:$$\begin{array}{c}\hfill {y}_{f}^{l}={\sigma}^{l}\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}\left(\sum _{g=1}^{l-1}{\mathcal{P}}_{f,g}^{l}\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}{y}_{g}^{l-1}\right)\end{array}$$
- -
- ${\sigma}^{l}$ represents the nonlinear activation function (i.e., ReLU); and
- -
- ${\mathcal{P}}_{f,g}^{l}\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}{y}_{g}^{l-1}$ indicates the graph structure relating the gth input ${y}_{g}^{l-1}$ to the fth output ${y}_{f}^{l}$.

Now, we simply combine two cases to model the mechanism of a convolutional network applied to a non-Euclidean graph in each time slot: the case in which edges vary, and that in which nodes vary. This can be combined into a single expression to describe ${\mathcal{P}}_{f,g}^{l}$ given by Equation (5):$$\begin{array}{c}\hfill {\mathcal{P}}_{f,g}^{l}\left(S\right)=\sum _{k=1}^{K}{{\rm Y}}_{f,g}^{l,(k:1)}+\sum _{k=0}^{K}\left(\prod _{m=0}^{k}{{\rm Y}}_{d}^{\left(m\right)}+{\varphi}_{k}\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}{\tau}^{k}\right)\end{array}$$- -
- ${\sum}_{k=1}^{K}{{\rm Y}}_{f,g}^{l,(k:1)}$ represents the edge-varying case, in which
- *
- ${{\rm Y}}_{f,g}^{l,(k:1)}$ acts as a shift operator, and therefore represents a learning paradigm for data embedded within complex graphs, whose weights are known to some degree of ambiguity, are only partially known, or are unknown.

- -
- ${\sum}_{k=0}^{K}\left({\prod}_{m=0}^{k}{{\rm Y}}_{d}^{\left(m\right)}+{\varphi}_{k}\phantom{\rule{0.166667em}{0ex}}\xb7\phantom{\rule{0.166667em}{0ex}}{\tau}^{k}\right)$ represents the node-varying case, in which
- *
- $d\subset E$ is a special set of nodes (i.e., nodes with a degree above a certain threshold, nodes with a certain level of hierarchy in the organization, or any other relevant feature),
- *
- ${\varphi}_{k}\in {[0,1]}^{Nxd}$ is a binary matrix, and
- *
- ${\tau}^{k}$ is a vector describing the node parameters in d.

- Pooling in non-Euclidean complex-networked cyber–physical graph time-dependent signals.As introduced earlier, downsampling pooling layers in classical deep-learning architectures that extract information from Euclidean domains such as speech, images, or videos typically report the maximal output within rectangular proximity [93]. In this way, it is possible to extract local characteristics that are shared by other areas of the images, thus considerably reducing the number of parameters that the deep network has to learn without sacrificing its learning capacity. Pooling can be described as a progressive coarsening of the graph. A simple way to do this is to collapse edges and reduce the size of the graph through a standard max-pooling operation on the nodes by just taking the maximum of each one of the feature tensors on each of the nodes being coarsened. This can be represented as a binary-tree structure of node indices. These pooling modules on graphs can be inserted between the convolutional modules in order to extract high-level graph representations, and thus be able to perform effective graph classification.Some alternatives in this field have not been to try to pool the whole network, but different hierarchies of the complex network in order to be able to learn which node groups have similar characteristics [94]. Once these groups are learned, clusters are made, and network pooling is carried out as described above or with an alternative method. This process is repeated for each of the network layers; thus, its classification is obtained. This presupposes, however, prior knowledge of the network structure.The extraction of shared local characteristics is not possible through this method in time-varying non-Euclidean domains, i.e., complex-networked cyber–physical graphs, because no stationarity or shift invariance can be found within these domains. Wu et al. [95], and Lee et al. [96] provided state-of-the-art surveying overview of this interesting open research question.

## 5. Conclusions and Management Implications

- Graphwise classification. The classification of complex cyber–physical graphs by deep lean learning, thus creating product families and allowing automated decision making in real time in which products are developed, produced, and channeled to the final customer.
- Vertexwise classification. The classification of certain crucial nodes in the value-creation process by means of deep-lean-learning models that allows an improvement of organizational design to assure an increase of overall process performance.
- Graph dynamics. Learning complex-networked cyber–physical graph dynamics is of great interest when dealing with change management within non-Euclidean sociotechnical systems.
- Known vs. unknown domains. The learning, generation, and semisupervised design of value streams by learning the most suitable complex cyber–physical graphs for certain types of products, thus potentially generating high customization with high efficiency and effectiveness in resource use.

## Author Contributions

## Funding

## Acknowledgments

## Conflicts of Interest

## Abbreviations

IIoT | Industrial Internet of Things |

## References

- Reinsel, D.; Gantz, J.; Rydning, J. The Digitization of the World. From Edge to Core. 2018. Available online: https://www.seagate.com/files/www-content/our-story/trends/files/idc-seagate-dataage-whitepaper.pdf (accessed on 30 January 2020).
- Ahmed, E.; Yaqoob, I.; Hashem, I.A.T.; Khan, I.; Ahmed, A.I.A.; Imran, M.; Vasilakos, A.V. The role of big data analytics in Internet of Things. Comput. Netw.
**2017**, 129, 459–471. [Google Scholar] [CrossRef] - Bhattacharjya, A.; Zhong, X.; Wang, J.; Li, X. Security Challenges and Concerns of Internet of Things (IoT). In Cyber-Physical Systems: Architecture, Security and Application; Guo, S., Zeng, D., Eds.; Springer International Publishing: Cham, Switzerland, 2019; pp. 153–185. [Google Scholar] [CrossRef]
- Jiang, P.; Leng, J.; Ding, K.; Gu, P.; Koren, Y. Social manufacturing as a sustainable paradigm for mass individualization. Proc. Inst. Mech. Eng. Part B J. Eng. Manuf.
**2016**, 230, 1961–1968. [Google Scholar] [CrossRef] - Goodfellow, I.; Bengio, Y.; Courville, A. Deep Learning, 1st ed.; Adaptive Computation and Machine Learning Series; The MIT Press: Cambridge, MA, USA, 2016. [Google Scholar]
- Hermann, M.; Pentek, T.; Otto, B. Design Principles for Industrie 4.0 Scenarios. In Proceedings of the 2016 49th Hawaii International Conference on System Sciences (HICSS), Koloa, HI, USA, 5–8 January 2016; pp. 3928–3937. [Google Scholar] [CrossRef] [Green Version]
- Mosterman, P.J.; Zander, J. Industry 4.0 as a Cyber-Physical System study. Softw. Syst. Model.
**2016**, 15, 17–29. [Google Scholar] [CrossRef] - Shah, R.; Ward, P.T. Lean manufacturing: Context, practice bundles, and performance. J. Oper. Manag.
**2003**, 21, 129–149. [Google Scholar] [CrossRef] - Villalba-Diez, J.; Ordieres-Mere, J.; Chudzick, H.; Lopez-Rojo, P. NEMAWASHI: Attaining Value Stream alignment within Complex Organizational Networks. Procedia CIRP
**2015**, 37, 134–139. [Google Scholar] [CrossRef] [Green Version] - Villalba-Diez, J.; Ordieres-Mere, J. Improving manufacturing operational performance by standardizing process management. Trans. Eng. Manag.
**2015**, 62, 351–360. [Google Scholar] [CrossRef] - Villalba-Diez, J.; Ordieres-Meré, J. Strategic Lean Organizational Design: Towards Lean World-Small World Configurations through Discrete Dynamic Organizational Motifs. Math. Probl. Eng.
**2016**, 2016, 1–10. [Google Scholar] [CrossRef] - Villalba-Diez, J.; Ordieres-Mere, J.; Rubio-Valdehita, S. Lean Learning Patterns. (CPD)nA vs. KATA. Procedia CIRP
**2016**, 54, 147–151. [Google Scholar] [CrossRef] [Green Version] - Villalba-Diez, J.; Ordieres-Meré, J.; Molina, M.; Rossner, M.; Lay, M. Lean Dendrochronology: Complexity Reduction by Representation of KPI Dynamics Looking at Strategic Organizational Design. Manag. Prod. Eng. Rev.
**2018**, 9, 3–9. [Google Scholar] [CrossRef] - Villalba-Diez, J. The Lean Brain Theory. Complex Networked Lean Strategic Organizational Design; CRC Press: Boca Raton, FL, USA, 2017. [Google Scholar]
- Villalba-Diez, J. The HOSHIN KANRI FOREST. Lean Strategic Organizational Design, 1st ed.; CRC Press: Boca Raton, FL, USA, 2017. [Google Scholar]
- Cross, R.L.; Singer, J.; Colella, S.; Thomas, R.J.; Silverstone, Y. (Eds.) The Organizational Network Fieldbook: Best Practices, Techniques and Exercises to Drive Organizational Innovation and Performance, 1st ed.; Jossey-Bass: San Francisco, CA, USA, 2010. [Google Scholar]
- Jabeur, N.; Sahli, N.; Zeadally, S. Enabling Cyber Physical Systems with Wireless Sensor Networking Technologies, Multiagent System Paradigm, and Natural Ecosystems. Mob. Inf. Syst.
**2015**, 2015, 15. [Google Scholar] [CrossRef] [Green Version] - Saleh, M.; Esa, Y.; Mohamed, A. Applications of Complex Network Analysis in Electric Power Systems. Energies
**2018**, 11, 1381. [Google Scholar] [CrossRef] [Green Version] - Barabási, A.-L. Network Science; Cambridge University Press: Cambridge, UK, 2016. [Google Scholar]
- Wang, Y.; Yuan, Y.; Ma, Y.; Wang, G. Time-Dependent Graphs: Definitions, Applications, and Algorithms. Data Sci. Eng.
**2019**. [Google Scholar] [CrossRef] [Green Version] - Suzaki, K. The New Shopfloor Management; The Free Press: New York, NY, USA, 1993. [Google Scholar]
- De Leeuw, S.; van der Berg, J.P. Improving operational performance by influencing shopfloor behavior via performance management practices. J. Oper. Manag.
**2011**, 29, 224–235. [Google Scholar] [CrossRef] - Villalba-Diez, J.; Ordieres-Meré, J.; Nuber, G. The HOSHIN KANRI TREE. Cross-Plant Lean Shopfloor Management. Procedia CIRP
**2015**, 32, 150–155. [Google Scholar] [CrossRef] [Green Version] - Villalba-Diez, J.; Zheng, X.; Schmidt, D.; Molina, M. Characterization of Industry 4.0 Lean Management Problem-Solving Behavioral Patterns Using EEG Sensors and Deep Learning. Sensors
**2019**, 19, 2841. [Google Scholar] [CrossRef] [Green Version] - Villalba-Diez, J.; Schmidt, D.; Gevers, R.; Ordieres-Meré, J.; Buchwitz, M.; Wellbrock, W. Deep Learning for Industrial Computer Vision Quality Control in the Printing Industry 4.0. Sensors
**2019**, 19, 3987. [Google Scholar] [CrossRef] [Green Version] - Imai, M. Gemba Kaizen: A Commonsense Approach to a Continuous Improvement Strategy, 2nd ed.; McGraw-Hill Professional: New York, NY, USA, 2012. [Google Scholar]
- Stock, T.; Seliger, G. Opportunities of Sustainable Manufacturing in Industry 4.0. Procedia CIRP
**2016**, 40, 536–541. [Google Scholar] [CrossRef] [Green Version] - Takeda, H. Intelligent Automation Textbook; Nikkan Kogyo Shimbun: Tokyo, Japan, 2009. [Google Scholar]
- Francis, J.; Bian, L. Deep Learning for Distortion Prediction in Laser-Based Additive Manufacturing using Big Data. Manuf. Lett.
**2019**, 20, 10–14. [Google Scholar] [CrossRef] - Li, B.H.; Hou, B.C.; Yu, W.T.; Lu, X.B.; Yang, C.W. Applications of artificial intelligence in intelligent manufacturing: A review. Front. Inf. Technol. Electron. Eng.
**2017**, 18, 86–96. [Google Scholar] [CrossRef] - Aazam, M.; Zeadally, S.; Harras, K.A. Deploying Fog Computing in Industrial Internet of Things and Industry 4.0. IEEE Trans. Ind. Inform.
**2018**, 14, 4674–4682. [Google Scholar] [CrossRef] - Tao, F.; Qi, Q.; Liu, A.; Kusiak, A. Data-driven smart manufacturing. J. Manuf. Syst.
**2018**, 48, 157–169. [Google Scholar] [CrossRef] - Mushtaq, A.; Haq, I.U. Implications of Blockchain in Industry 4.O. In Proceedings of the 2019 International Conference on Engineering and Emerging Technologies (ICEET), Lahore, Pakistan, 21–22 February 2019; pp. 1–5. [Google Scholar] [CrossRef]
- Shevchik, S.A.; Masinelli, G.; Kenel, C.; Leinenbach, C.; Wasmer, K. Deep Learning for In Situ and Real-Time Quality Monitoring in Additive Manufacturing Using Acoustic Emission. IEEE Trans. Ind. Inform.
**2019**, 15, 5194–5203. [Google Scholar] [CrossRef] - Al-Jaroodi, J.; Mohamed, N. Blockchain in Industries: A Survey. IEEE Access
**2019**, 7, 36500–36515. [Google Scholar] [CrossRef] - Sun, S.; Zheng, X.; Villalba-Diez, J.; Ordieres-Meré, J. Indoor Air-Quality Data-Monitoring System: Long-Term Monitoring Benefits. Sensors
**2019**, 19, 4157. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Rother, M. Toyota Kata. Managing People for Improvement, Adaptiveness, and Superior Results; McGraw-Hill: New York, NY, USA, 2010. [Google Scholar]
- Birkel, S.H.; Veile, W.J.; Müller, M.J.; Hartmann, E.; Voigt, K.I. Development of a Risk Framework for Industry 4.0 in the Context of Sustainability for Established Manufacturers. Sustainability
**2019**, 11, 384. [Google Scholar] [CrossRef] [Green Version] - Takeda, H. LCIA—Low Cost Intelligent Automation, 3rd ed.; Finanzbuch Verlag GmbH: Munich, Germany, 2011. [Google Scholar]
- Davis, J.; Edgar, T.; Porter, J.; Bernaden, J.; Sarli, M. Smart manufacturing, manufacturing intelligence and demand-dynamic performance. Comput. Chem. Eng.
**2012**, 47, 145–156. [Google Scholar] [CrossRef] - Gómez, A.; Cuiñas, D.; Catalá, P.; Xin, L.; Li, W.; Conway, S.; Lack, D. Use of Single Board Computers as Smart Sensors in the Manufacturing Industry. MESIC Manuf. Eng. Soc. Int. Conf.
**2015**, 132, 153–159. [Google Scholar] [CrossRef] [Green Version] - Culot, G.; Orzes, G.; Sartor, M. Integration and Scale in the Context of Industry 4.0: The Evolving Shapes of Manufacturing Value Chains. IEEE Eng. Manag. Rev.
**2019**, 47, 45–51. [Google Scholar] [CrossRef] - Jiménez, P.; Villalba-Díez, J.; Ordieres-Meré, J. HOSHIN KANRI Visualization with Neo4j. Empowering Leaders to Operationalize Lean Structural Networks. Procedia CIRP
**2016**, 55, 284–289. [Google Scholar] - Wang, J.; Ma, Y.; Zhang, L.; Gao, R.; Wu, D. Deep learning for smart manufacturing: Methods and applications. J. Manuf. Syst.
**2018**. [Google Scholar] [CrossRef] - Jang, S.H.; Guejong, J.; Jeong, J.; Sangmin, B. Fog Computing Architecture Based Blockchain for Industrial IoT. Computational Science—ICCS 2019; Rodrigues, J.M.F., Cardoso, P.J.S., Monteiro, J., Lam, R., Krzhizhanovskaya, V.V., Lees, M.H., Dongarra, J.J., Sloot, P.M., Eds.; Springer International Publishing: Cham, Switzerland, 2019; pp. 593–606. [Google Scholar]
- Ordieres-Meré, J.; Villalba-Diez, J.; Zheng, X. Challenges and Opportunities for Publishing IIoT Data in Manufacturing as a Service Business; Cyber Physical Manufacturing; Procedia Manufacturing; Elsevier: Chicago, IL, USA, 2019. [Google Scholar]
- Villalba-Diez, J.; DeSanctis, I.; Ordieres-Meré, J.; Ciarapica, F. Lean Structural Network Resilience. In Complex Networks and Its Applications VI: Proceedings of Complex Networks 2017 (The Sixth International Conference on Complex Networks and Their Applications); Cherifi, C., Cherifi, H., Karsai, M., Musolesi, M., Eds.; Number 689 in Studies in Computational Intelligence; Springer International Publishing: Lyon, France, 2017; pp. 609–619. [Google Scholar]
- Davies, R.; Coole, T.; Smith, A. Review of Socio-technical Considerations to Ensure Successful Implementation of Industry 4.0. Procedia Manuf.
**2017**, 11, 1288–1295. [Google Scholar] [CrossRef] - Kumar, K.; Zindani, D.; Davim, J.P. Socio-technical Considerations. In Industry 4.0: Developments towards the Fourth Industrial Revolution; Kumar, K., Zindani, D., Davim, J.P., Eds.; Springer: Singapore, 2019; pp. 43–51. [Google Scholar] [CrossRef]
- Womack, J.P.; Jones, D.T. Introduction. In Lean Thinking, 2nd ed.; Simon and Schuster: New York, NY, USA, 2003; pp. 15–28. [Google Scholar]
- Toyota Motor Corporation. Toyota Motor Corporation. Sustainability Report 2013; Sustainability Report; Toyota Motor Corporation: Tokyo, Jaoan, 2014. [Google Scholar]
- Burton, R.M.; Øbel, B.; Håkonsson, D.D. Organizational Design: A Step-by-Step Approach, 3rd ed.; Cambridge University Press: Cambridge, UK, 2015. [Google Scholar]
- Covey, S.R. The 8th Habit. From Effectiveness to Greatness; Free Press: New York, NY, USA, 2004. [Google Scholar]
- Rabelo, R.J.; Zambiasi, S.P.; Romero, D. Collaborative Softbots: Enhancing Operational Excellence in Systems of Cyber-Physical Systems. In Collaborative Networks and Digital Transformation; Camarinha-Matos, L.M., Afsarmanesh, H., Antonelli, D., Eds.; Springer International Publishing: Cham, Switzerland, 2019; pp. 55–68. [Google Scholar]
- Romero, D.; Wuest, T.; Stahre, J.; Gorecky, D. Social Factory Architecture: Social Networking Services and Production Scenarios Through the Social Internet of Things, Services and People for the Social Operator 4.0. In Advances in Production Management Systems. The Path to Intelligent, Collaborative and Sustainable Manufacturing; Lödding, H., Riedel, R., Thoben, K.D., von Cieminski, G., Kiritsis, D., Eds.; Springer International Publishing: Cham, Switzerland, 2017; pp. 265–273. [Google Scholar]
- Wang, F.; Shang, X.; Qin, R.; Xiong, G.; Nyberg, T.R. Social Manufacturing: A Paradigm Shift for Smart Prosumers in the Era of Societies 5.0. IEEE Trans. Comp. Social Syst.
**2019**, 6, 822–829. [Google Scholar] [CrossRef] - Guo, W.; Jiang, P. Product Service Systems for Social Manufacturing: A new service system with multi-provider. IFAC-PapersOnLine
**2019**, 52, 749–754. [Google Scholar] [CrossRef] - Lee, J.; Bagheri, B.; Kao, H.A. A Cyber-Physical Systems architecture for Industry 4.0-based manufacturing systems. Manuf. Lett.
**2015**, 3, 18–23. [Google Scholar] [CrossRef] - Wang, L.; Toerngren, M.; Onori, M. Current status and advancement of cyber-physical systems in manufacturing. J. Manuf. Syst.
**2015**, 37, 517–527. [Google Scholar] [CrossRef] - Sisinni, E.; Saifullah, A.; Han, S.; Jennehag, U.; Gidlund, M. Industrial Internet of Things: Challenges, Opportunities, and Directions. IEEE Trans. Ind. Inform.
**2018**, 14, 4724–4734. [Google Scholar] [CrossRef] - Zheng, P.; Wang, H.; Sang, Z.; Zhong, R.Y.; Liu, Y.; Liu, C.; Mubarok, K.; Yu, S.; Xu, X. Smart manufacturing systems for Industry 4.0: Conceptual framework, scenarios, and future perspectives. Front. Mech. Eng.
**2018**, 13, 137–150. [Google Scholar] [CrossRef] - Lu, Y.; Xu, X. Cloud-based manufacturing equipment and big data analytics to enable on-demand manufacturing services. Robot. Comput. Integr. Manuf.
**2019**, 57, 92–102. [Google Scholar] [CrossRef] - Kiel, D.; MÜller, J.M.; Arnold, C.; Voigt, K.I. Sustainable Industrial Value Creation: Benefits and Challenges of Industry 4.0. Int. J. Innov. Manag.
**2017**, 21, 1740015. [Google Scholar] [CrossRef] - Stock, T.; Obenaus, M.; Kunz, S.; Kohl, H. Industry 4.0 as enabler for a sustainable development: A qualitative assessment of its ecological and social potential. Process Saf. Environ. Prot.
**2018**, 118, 254–267. [Google Scholar] [CrossRef] - Shang, X.; Shen, Z.; Xiong, G.; Wang, F.Y.; Liu, S.; Nyberg, T.R.; Wu, H.; Guo, C. Moving from mass customization to social manufacturing: A footwear industry case study. Int. J. Comput. Integr. Manuf.
**2019**, 32, 194–205. [Google Scholar] [CrossRef] - Leng, J.; Jiang, P. A deep learning approach for relationship extraction from interaction context in social manufacturing paradigm. Knowl. Based Syst.
**2016**, 100, 188–199. [Google Scholar] [CrossRef] - Leng, J.; Chen, Q.; Mao, N.; Jiang, P. Combining granular computing technique with deep learning for service planning under social manufacturing contexts. Knowl. Based Syst.
**2018**, 143, 295–306. [Google Scholar] [CrossRef] - Subakti, H.; Jiang, J.-R. Indoor Augmented Reality Using Deep Learning for Industry 4.0 Smart Factories. In Proceedings of the 2018 IEEE 42nd Annual Computer Software and Applications Conference (COMPSAC), Tokyo, Japan, 23–27 July 2018; Volume 2, pp. 63–68. [Google Scholar] [CrossRef]
- Tao, F.; Qi, Q.; Wang, L.; Nee, A.Y.C. Digital Twins and Cyber–Physical Systems toward Smart Manufacturing and Industry 4.0: Correlation and Comparison. Engineering
**2019**, 5, 653–661. [Google Scholar] [CrossRef] - Bronstein, M.M.; Bruna, J.; LeCun, Y.; Szlam, A.; Vandergheynst, P. Geometric Deep Learning: Going beyond Euclidean data. IEEE Signal Process. Mag.
**2017**, 34, 18–42. [Google Scholar] [CrossRef] [Green Version] - Ma, Y.; Fu, Y. Manifold Learning Theory and Applications; CRC Press: Boca Raton, FL, USA, 2011. [Google Scholar]
- Keller, M. Curvature, Geometry and Spectral Properties of Planar Graphs. Discret. Comput. Geom.
**2011**, 46, 500–525. [Google Scholar] [CrossRef] [Green Version] - Wu, Z.; Menichetti, G.; Rahmede, C.; Bianconi, G. Emergent complex network geometry. Sci. Rep.
**2015**, 5, 10073. [Google Scholar] [CrossRef] [Green Version] - Bianconi, G.; Rahmede, C.; Wu, Z. Complex quantum network geometries: Evolution and phase transitions. Phys. Rev. E Stat. Nonlinear Soft Matter Phys.
**2015**, 92, 022815. [Google Scholar] [CrossRef] [Green Version] - Bianconi, G.; Rahmede, C. Complex Quantum Network Manifolds in Dimension d > 2 are Scale-Free. Sci. Rep.
**2015**, 5, 13979. [Google Scholar] [CrossRef] [Green Version] - Zhang, Z.; Cui, P.; Zhu, W. Deep Learning on Graphs: A Survey. arXiv
**2018**, arXiv:1812.04202. [Google Scholar] - Duvenaud, D.; Maclaurin, D.; Aguilera-Iparraguirre, J.; Gómez-Bombarelli, R.; Hirzel, T.; Aspuru-Guzik, A.; Adams, R.P. Convolutional Networks on Graphs for Learning Molecular Fingerprints. arXiv
**2015**, arXiv:1509.09292. [Google Scholar] - Krizhevsky, A.; Sutskever, L.; Hinton, G. Imagenet classification with deep convolutional neural networks. In Proceedings of the 25th International Conference on Neural Information Processing Systems, Lake Tahoe, CA, USA, 3–8 December 2012; pp. 1106–1114. [Google Scholar]
- Chen, L.C.; Barron, J.T.; Papandreou, G.; Murphy, K.; Yuille, A.L. Semantic Image Segmentation with Task-Specific Edge Detection Using CNNs and a Discriminatively Trained Domain Transform. arXiv
**2015**, arXiv:1511.03328. [Google Scholar] - Cao, S.; Lu, W.; Xu, Q. Deep Neural Networks for Learning Graph Representations. In Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence, Phoenix, AZ, USA, 12–17 February 2016; pp. 1145–1152. [Google Scholar]
- Shuman, D.; Narang, S.; Frossard, P.; Ortega, A.; Vandergheynst, P. The emerging field of signal processing on graphs: Extending high-dimensional data analysis to networks and other irregular domains. IEEE Signal Process. Mag.
**2013**, 30, 83–98. [Google Scholar] [CrossRef] [Green Version] - Henaff, M.; Bruna, J.; LeCun, Y. Deep convolutional networks on graph-structured data. arXiv
**2015**, arXiv:1506.05163. [Google Scholar] - Defferrard, M.; Bresson, X.; Vandergheynst, P. Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering. arXiv
**2016**, arXiv:1606.09375. [Google Scholar] - Kipf, T.N.; Welling, M. Semi-Supervised Classification with Graph Convolutional Networks. arXiv
**2016**, arXiv:1609.02907. [Google Scholar] - Qiu, J.; Tang, J.; Ma, H.; Dong, Y.; Wang, K.; Tang, J. DeepInf: Social Influence Prediction with Deep Learning. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, London, UK, 19–23 August 2018; ACM: New York, NY, USA, 2018; pp. 2110–2119. [Google Scholar] [CrossRef] [Green Version]
- Garcia, V.; Bruna, J. Few-Shot Learning with Graph Neural Networks. arXiv
**2017**, arXiv:1711.04043. [Google Scholar] - Jain, A.; Zamir, A.R.; Savarese, S.; Saxena, A. Structural-RNN: Deep Learning on Spatio-Temporal Graphs. arXiv
**2015**, arXiv:1511.05298. [Google Scholar] - Marino, K.; Salakhutdinov, R.; Gupta, A. The More You Know: Using Knowledge Graphs for Image Classification. arXiv
**2016**, arXiv:1612.04844. [Google Scholar] - Sandryhaila, A.; Moura, J.M.F. Discrete Signal Processing on Graphs. IEEE Trans. Signal Process.
**2013**, 61, 1644–1656. [Google Scholar] [CrossRef] [Green Version] - Harris, K.D.; Aravkin, A.; Rao, R.; Brunton, B.W. Time-varying Autoregression with Low Rank Tensors. arXiv
**2019**, arXiv:1905.08389. [Google Scholar] - Sciriha, I. A characterization of singular graphs. Electron. J. Linear Algebra
**2007**, 16, 451–462. [Google Scholar] [CrossRef] [Green Version] - Ortega, A.; Frossard, P.; Kovačević, J.; Moura, J.M.; Vandergheynst, P. Graph Signal Processing: Overview, Challenges, and Applications. Proc. IEEE
**2018**, 106, 808–828. [Google Scholar] [CrossRef] [Green Version] - Zhou, Y.T.; Chellappa, R. Computation of optical flow using a neural network. In Proceedings of the IEEE 1988 International Conference on Neural Networks, San Diego, CA, USA, 24–27 July 1988; Volume 2, pp. 71–78. [Google Scholar] [CrossRef]
- Ying, Z.; You, J.; Morris, C.; Ren, X.; Hamilton, W.; Leskovec, J. Hierarchical Graph Representation Learning with Differentiable Pooling. In Advances in Neural Information Processing Systems 31; Bengio, S., Wallach, H., Larochelle, H., Grauman, K., Cesa-Bianchi, N., Garnett, R., Eds.; Curran Associates, Inc.: Red Hook, NY, USA, 2018; pp. 4800–4810. [Google Scholar]
- Wu, Z.; Pan, S.; Chen, F.; Long, G.; Zhang, C.; Yu, P.S. A Comprehensive Survey on Graph Neural Networks. arXiv
**2019**, arXiv:1901.00596. [Google Scholar] - Lee, J.; Lee, I.; Kang, J. Self-Attention Graph Pooling. In Proceedings of the 36th International Conference on Machine Learning, Long Beach, CA, USA, 9–15 June 2019; Chaudhuri, K., Salakhutdinov, R., Eds.; PMLR: Long Beach, CA, USA, 2019; Volume 97, pp. 3734–3743. [Google Scholar]

**Figure 1.**Macroscopic, mesoscopic, and microscopic levels of organizational sociotechnical complexity.

Social | Technical | Socio Technical | |
---|---|---|---|

Micro | Imai, 2012 [26]; Stock and Seliger, 2016 [27] | Takeda, 2009 [28]; Francis and Bian, 2019 [29]; Jabeur et al., 2015 [17]; Li et al., 2017 [30]; Aazam et al., 2018 [31]; Tao et al., 2018 [32]; Mushtaq and Haq, 2019 [33]; Shevchik et al., 2019 [34]; Al-Jaroodi and Mohamed, 2019 [35]; Sun et al., 2019 [36] | Villalba-Diez et al., 2015 [23]; Villalba-Diez et al., 2019 [24]. |

Meso | Rother, 2010 [37]; Villalba et al., 2018 [13]; Birkel et al., 2019 [38] | Takeda, 2011 [39]; Davis et al., 2012 [40]; Gomez et al., 2015 [41]; Culot, 2019 [42]; Jimenez et al., 2016 [43]; Wang et al., 2018 [44]; Villalba-Diez et al., 2019 [25]; Jang et al., 2019 [45]; Ordieres-Mere et al., 2019 [46] | Villalba-Diez and Ordieres-Mere, 2015 [10]; Villalba-Diez et al., 2015 [9]; Villalba-Diez and Ordieres-Mere, 2016 [11]; Villalba-Diez et al., 2017 [47]; Davies et al., 2017 [48]; Kumar et al., 2019 [49]. |

Macro | Womack and Jones, 2003 [50]; Toyota, 2014 [51]; Burton et al., 2015 [52]; Covey, 2004 [53]; Rabelo et al., 2019 [54]; Romero et al., 2017 [55]; Wang et al., 2019 [56]; Guo and Jyang, 2019 [57] | Lee et al., 2015 [58]; Wang et al., 2015 [59]; Goodfellow et al., 2016 [5]; Sisini et al., 2018 [60]; Zheng et al., 2018a [61]; Lu and Xu, 2019 [62] | Stock and Seliger, 2016 [27]; Villalba-Diez, 2017 [15]; Villalba-Diez, 2017 [14]; Kiel et al., 2017 [63]; Stock et al., 2016 [64]; Shang et al., 2019 [65]. |

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Villalba-Díez, J.; Molina, M.; Ordieres-Meré, J.; Sun, S.; Schmidt, D.; Wellbrock, W.
Geometric Deep Lean Learning: Deep Learning in Industry 4.0 Cyber–Physical Complex Networks. *Sensors* **2020**, *20*, 763.
https://doi.org/10.3390/s20030763

**AMA Style**

Villalba-Díez J, Molina M, Ordieres-Meré J, Sun S, Schmidt D, Wellbrock W.
Geometric Deep Lean Learning: Deep Learning in Industry 4.0 Cyber–Physical Complex Networks. *Sensors*. 2020; 20(3):763.
https://doi.org/10.3390/s20030763

**Chicago/Turabian Style**

Villalba-Díez, Javier, Martin Molina, Joaquín Ordieres-Meré, Shengjing Sun, Daniel Schmidt, and Wanja Wellbrock.
2020. "Geometric Deep Lean Learning: Deep Learning in Industry 4.0 Cyber–Physical Complex Networks" *Sensors* 20, no. 3: 763.
https://doi.org/10.3390/s20030763