Actionable Cognitive Twins are the next generation Digital Twins enhanced with cognitive capabilities through a knowledge graph and artificial intelligence models that provide insights and decision-making options to the users. The knowledge graph des
cribes the domain-specific knowledge regarding entities and interrelationships related to a manufacturing setting. It also contains information on possible decision-making options that can assist decision-makers, such as planners or logisticians. In this paper, we propose a knowledge graph modeling approach to construct actionable cognitive twins for capturing specific knowledge related to demand forecasting and production planning in a manufacturing plant. The knowledge graph provides semantic descriptions and contextualization of the production lines and processes, including data identification and simulation or artificial intelligence algorithms and forecasts used to support them. Such semantics provide ground for inferencing, relating different knowledge types: creative, deductive, definitional, and inductive. To develop the knowledge graph models for describing the use case completely, systems thinking approach is proposed to design and verify the ontology, develop a knowledge graph and build an actionable cognitive twin. Finally, we evaluate our approach in two use cases developed for a European original equipment manufacturer related to the automotive industry as part of the European Horizon 2020 project FACTLOG.
Although there exist several libraries for deep learning on graphs, they are aiming at implementing basic operations for graph deep learning. In the research community, implementing and benchmarking various advanced tasks are still painful and time-c
onsuming with existing libraries. To facilitate graph deep learning research, we introduce DIG: Dive into Graphs, a research-oriented library that integrates unified and extensible implementations of common graph deep learning algorithms for several advanced tasks. Currently, we consider graph generation, self-supervised learning on graphs, explainability of graph neural networks, and deep learning on 3D graphs. For each direction, we provide unified implementations of data interfaces, common algorithms, and evaluation metrics. Altogether, DIG is an extensible, open-source, and turnkey library for researchers to develop new methods and effortlessly compare with common baselines using widely used datasets and evaluation metrics. Source code is available at https://github.com/divelab/DIG.
We investigate the optimal portfolio deleveraging (OPD) problem with permanent and temporary price impacts, where the objective is to maximize equity while meeting a prescribed debt/equity requirement. We take the real situation with cross impact amo
ng different assets into consideration. The resulting problem is, however, a non-convex quadratic program with a quadratic constraint and a box constraint, which is known to be NP-hard. In this paper, we first develop a successive convex optimization (SCO) approach for solving the OPD problem and show that the SCO algorithm converges to a KKT point of its transformed problem. Second, we propose an effective global algorithm for the OPD problem, which integrates the SCO method, simple convex relaxation and a branch-and-bound framework, to identify a global optimal solution to the OPD problem within a pre-specified $epsilon$-tolerance. We establish the global convergence of our algorithm and estimate its complexity. We also conduct numerical experiments to demonstrate the effectiveness of our proposed algorithms with both the real data and the randomly generated medium- and large-scale OPD problem instances.
Properties of molecules are indicative of their functions and thus are useful in many applications. With the advances of deep learning methods, computational approaches for predicting molecular properties are gaining increasing momentum. However, the
re lacks customized and advanced methods and comprehensive tools for this task currently. Here we develop a suite of comprehensive machine learning methods and tools spanning different computational models, molecular representations, and loss functions for molecular property prediction and drug discovery. Specifically, we represent molecules as both graphs and sequences. Built on these representations, we develop novel deep models for learning from molecular graphs and sequences. In order to learn effectively from highly imbalanced datasets, we develop advanced loss functions that optimize areas under precision-recall curves. Altogether, our work not only serves as a comprehensive tool, but also contributes towards developing novel and advanced graph and sequence learning methodologies. Results on both online and offline antibiotics discovery and molecular property prediction tasks show that our methods achieve consistent improvements over prior methods. In particular, our methods achieve #1 ranking in terms of both ROC-AUC and PRC-AUC on the AI Cures Open Challenge for drug discovery related to COVID-19. Our software is released as part of the MoleculeX library under AdvProp.
Model-based Systems Engineering (MBSE) has been widely utilized to formalize system artifacts and facilitate their development throughout the entire lifecycle. During complex system development, MBSE models need to be frequently exchanged across stak
eholders. Concerns about data security and tampering using traditional data exchange approaches obstruct the construction of a reliable marketplace for digital assets. The emerging Distributed Ledger Technology (DLT), represented by blockchain, provides a novel solution for this purpose owing to its unique advantages such as tamper-resistant and decentralization. In this paper, we integrate MBSE approaches with DLT aiming to create a decentralized marketplace to facilitate the exchange of digital engineering assets (DEAs). We first define DEAs from perspectives of digital engineering objects, development processes and system architectures. Based on this definition, the Graph-Object-Property-Point-Role-Relationship (GOPPRR) approach is used to formalize the DEAs. Then we propose a framework of a decentralized DEAs marketplace and specify the requirements, based on which we select a Directed Acyclic Graph (DAG) structured DLT solution. As a proof-of-concept, a prototype of the proposed DEAs marketplace is developed and a case study is conducted to verify its feasibility. The experiment results demonstrate that the proposed marketplace facilitates free DEAs exchange with a high level of security, efficiency and decentralization.
Cognitive Twins (CT) are proposed as Digital Twins (DT) with augmented semantic capabilities for identifying the dynamics of virtual model evolution, promoting the understanding of interrelationships between virtual models and enhancing the decision-
making based on DT. The CT ensures that assets of Internet of Things (IoT) systems are well-managed and concerns beyond technical stakeholders are addressed during IoT system development. In this paper, a Knowledge Graph (KG) centric framework is proposed to develop CT. Based on the framework, a future tool-chain is proposed to develop the CT for the initiatives of H2020 project FACTLOG. Based on the comparison between DT and CT, we infer the CT is a more comprehensive approach to support IoT-based systems development than DT.
Heat transport in one-dimensional (1D) momentum-conserving lattices is generally assumed to be anomalous, thus yielding a power-law divergence of thermal conductivity with system length. However, whether heat transport in two-dimensional (2D) system
is anomalous or not is still on debate because of the difficulties involved in experimental measurements or due to the insufficiently large simulation size. Here, we simulate energy and momentum diffusion in the 2D nonlinear lattices using the method of fluctuation correlation functions. Our simulations confirm that energy diffusion in the 2D momentum-conserving lattices is anomalous and can be well described by the L{e}vy-stable distribution. We also find that the disappear of side peaks of heat mode may suggest a weak coupling between heat mode and sound mode in the 2D nonlinear system. It is also observed that the harmonic interactions in the 2D nonlinear lattices can accelerate the energy diffusion. Contrary to the hypothesis of 1D system, we clarify that anomalous heat transport in the 2D momentum-conserving system cannot be corroborated by the momentum superdiffusion any more. Moreover, as is expected, lattices with a nonlinear on-site potential exhibit normal energy diffusion, independent of the dimension. Our findings offer some valuable insights into the mechanism of thermal transport in 2D system.
In mammography, the efficacy of computer-aided detection methods depends, in part, on the robust localisation of micro-calcifications ($mu$C). Currently, the most effective methods are based on three steps: 1) detection of individual $mu$C candidates
, 2) clustering of individual $mu$C candidates, and 3) classification of $mu$C clusters. Where the second step is motivated both to reduce the number of false positive detections from the first step and on the evidence that malignancy depends on a relatively large number of $mu$C detections within a certain area. In this paper, we propose a novel approach to $mu$C detection, consisting of the detection emph{and} classification of individual $mu$C candidates, using shape and appearance features, using a cascade of boosting classifiers. The final step in our approach then clusters the remaining individual $mu$C candidates. The main advantage of this approach lies in its ability to reject a significant number of false positive $mu$C candidates compared to previously proposed methods. Specifically, on the INbreast dataset, we show that our approach has a true positive rate (TPR) for individual $mu$Cs of 40% at one false positive per image (FPI) and a TPR of 80% at 10 FPI. These results are significantly more accurate than the current state of the art, which has a TPR of less than 1% at one FPI and a TPR of 10% at 10 FPI. Our results are competitive with the state of the art at the subsequent stage of detecting clusters of $mu$Cs.
In this paper we study the (equivariant) topological types of a class of 3-dimensional closed manifolds (i.e., 3-dimensional small covers), each of which admits a locally standard $(mathbb{Z}_2)^3$-action such that its orbit space is a simple convex
3-polytope. We introduce six equivariant operations on 3-dimensional small covers. These six operations are interesting because of their combinatorial natures. Then we show that each 3-dimensional small cover can be obtained from $mathbb{R}P^3$ and $S^1timesmathbb{R}P^2$ with certain $(mathbb{Z}_2)^3$-actions under these six operations. As an application, we classify all 3-dimensional small covers up to $({Bbb Z}_2)^3$-equivariant unoriented cobordism.
As a generalization of Davis-Januszkiewicz theory, there is an essential link between locally standard $(Z_2)^n$-actions (or $T^n$-actions) actions and nice manifolds with corners, so that a class of nicely behaved equivariant cut-and-paste operation
s on locally standard actions can be carried out in step on nice manifolds with corners. Based upon this, we investigate what kinds of closed manifolds admit locally standard $(Z_2)^n$-actions; especially for the 3-dimensional case. Suppose $M$ is an orientable closed connected 3-manifold. When $H_1(M;Z_2)=0$, it is shown that $M$ admits a locally standard $(Z_2)^3$-action if and only if $M$ is homeomorphic to a connected sum of 8 copies of some $Z_2$-homology sphere $N$, and if further assuming $M$ is irreducible, then $M$ must be homeomorphic to $S^3$. In addition, the argument is extended to rational homology 3-sphere $M$ with $H_1(M;Z_2) cong Z_2$ and an additional assumption that the $(Z_2)^3$-action has a fixed point.
