Showing papers on "Representation (systemics) published in 2022"

ChangeMask: Deep multi-task encoder-transformer-decoder architecture for semantic change detection

Abstract: Multi-temporal high spatial resolution earth observation makes it possible to detect complex urban land surface changes, which is a significant and challenging task in remote sensing communities. Previous works mainly focus on binary change detection (BCD) based on modern technologies, e.g., deep fully convolutional network (FCN), whereas the deep network architecture for semantic change detection (SCD) is insufficiently explored in current literature. In this paper, we propose a deep multi-task encoder-transformer-decoder architecture (ChangeMask) designed by exploring two important inductive biases: sematic-change causal relationship and temporal symmetry. ChangeMask decouples the SCD into a temporal-wise semantic segmentation and a BCD, and then integrates these two tasks into a general encoder-transformer-decoder framework. In the encoder part, we design a semantic-aware encoder to model the semantic-change causal relationship. This encoder is only used to learn semantic representation and then learn change representation from semantic representation via a later transformer module. In this way, change representation can constrain semantic representation during training, which introduces a regularization to reduce the risk of overfitting. To learn a robust change representation from semantic representation, we propose a temporal-symmetric transformer (TST) to guarantee temporal symmetry for change representation and keep it discriminative. Based on the above semantic representation and change representation, we adopt simple multi-task decoders to output semantic change map. Benefiting from the differentiable building blocks, ChangeMask can be trained by a multi-task loss function, which significantly simplifies the whole pipeline of applying ChangeMask. The comprehensive experimental results on two large-scale SCD datasets confirm the effectiveness and superiority of ChangeMask in SCD. Besides, to demonstrate the potential value in real-world applications, e.g., automatic urban analysis and decision-making, we deploy the ChangeMask to map a large geographic area covering 30 km2 with 300 million pixels. Code will be made available.

A Compact Representation of Measured BRDFs Using Neural Processes

Abstract: In this article, we introduce a compact representation for measured BRDFs by leveraging Neural Processes (NPs). Unlike prior methods that express those BRDFs as discrete high-dimensional matrices o...

Implications of Competitor Representation for Profit-Maximizing Design

Abstract: Design optimization studies that model competition with other products in the market often use a small set of products to represent all competitors. We investigate the effect of competitor product representation on profit-maximizing design solutions. Specifically, we study the implications of replacing a large set of disaggregated elemental competitor products with a subset of competitor products or composite products. We derive first-order optimality conditions and show that optimal design (but not price) is independent of competitors when using logit and nested logit models (where preferences are homogeneous). However, this relationship differs in the case of random-coefficients logit models (where preferences are heterogeneous), and we demonstrate that profit-maximizing design solutions using latent-class or mixed-logit models can (but need not always) depend on the representation of competing products. We discuss factors that affect the magnitude of the difference between models with elemental and composite representations of competitors, including preference heterogeneity, cost function curvature, and competitor set specification. We present correction factors that ensure models using subsets or composite representation of competitors have optimal design solutions that match those of disaggregated elemental models. While optimal designs using logit and nested logit models are not affected by ad hoc modeling decisions of competitor representation, the independence of optimal designs from competitors when using these models raises questions of when these models are appropriate to use.

Model-based clinical note entity recognition for rheumatoid arthritis using bidirectional encoder representation from transformers

Abstract: Background Rheumatoid arthritis (RA) is a disease of the immune system with a high rate of disability and there are a large amount of valuable disease diagnosis and treatment information in the clinical note of the electronic medical record. Artificial intelligence methods can be used to mine useful information in clinical notes effectively. This study aimed to develop an effective method to identify and classify medical entities in the clinical notes relating to RA and use the entity identification results in subsequent studies. Methods In this paper, we introduced the bidirectional encoder representation from transformers (BERT) pre-training model to enhance the semantic representation of word vectors. The generated word vectors were then inputted into the model, which is composed of traditional bidirectional long short-term memory neural networks and conditional random field machine learning algorithms for the named entity recognition of clinical notes to improve the model's effectiveness. The BERT method takes the combination of token embeddings, segment embeddings, and position embeddings as the model input and fine-tunes the model during training. Results Compared with the traditional Word2vec word vector model, the performance of the BERT pre-training model to obtain a word vector as model input was significantly improved. The best F1-score of the named entity recognition task after training using many rheumatoid arthritis clinical notes was 0.936. Conclusions This paper confirms the effectiveness of using an advanced artificial intelligence method to carry out named entity recognition tasks on a corpus of a large number of clinical notes; this application is promising in the medical setting. Moreover, the extraction of results in this study provides a lot of basic data for subsequent tasks, including relation extraction, medical knowledge graph construction, and disease reasoning.

Decision Premises, Learning and Organizational Resilience Addressing Novel Adversities

Abstract: Organizational resilience is traditionally associated with the ability to understand and to respond to the ongoing situation, even under unusual conditions. The capability to detect novel and unexpected situations plays a fundamental role in this process. Following (Simon‚ 1991), we believe that decision premises affect the problem representation and, ultimately, the possibility to detect, interpret and respond to novel situations, thus enhancing resilience. From this perspective, the ability to expand the perceptual limits of observation and to conceive a novel representation of the problem requires revising the initial decision premises. This theory of how organizations learn how to solve novel problems provides the foundation to introduce a role designed to legitimately challenge the decision premises and, ultimately, the spectrum of alternatives that are taken into consideration as possible solutions. To illustrate our proposal to increase organizational resilience, we introduce an exemplary real case stemming from the practice of the emergency management organizations under scrutiny of our research team; this case is reconstructed as a conversational narrative of the two key participants.

Efficient Representation for Electric Vehicle Charging Station Operations using Reinforcement Learning

Abstract: You are currently viewing a Conference Paper from the November 2021 Good Systems Network Digest.

The NEMESIS Macro-Econometric Model

Abstract: This Chapter presents the NEMESIS macro-econometric model. This model has been used for several ex-ante and ex-post evaluations of the macroeconomic impact of EU R&I policies. After a general overview of the model, a thorough description of the representation of innovation in the model is provided. As an example of its workings, an application to the interim evaluation of the Horizon 2020 programme is also provided.

Goal-directed generation of new molecules by AI methods

Abstract: Finding from scratch a new molecule with sought properties remains a challenge. In this chapter, we propose a presentation of the research challenges of molecular generation using methods from the artificial intelligence domain. This objective can be separated into three fundamental subproblems. The first problem is the representation of molecules. There is no universal molecular representation. It is, therefore, necessary to choose a representation adapted to the objective to be addressed. The second problem consists in guiding and evaluating generative methods. The role of the objective function is to orient the generation procedure toward solutions that address the problem. It is also possible to define assessment functions, whose purpose is to evaluate a set of solutions proposed by a method, beyond the criterion formulated by the objective function. The third problem consists in designing molecular generation methods. Many methods have been proposed in the last 30 years. We choose to present the methods according to the way they generate solutions and explore the molecular space. We highlight in this chapter the crucial role and interlocking nature of those three subproblems.

Application of Urquhart’s Representation of Lattices to Some Non–classical Logics

Abstract: Based on Alasdair Urquhart’s representation of not necessarily distributive bounded lattices we exhibit several discrete dualities in the spirit of the “duality via truth” concept by Orlowska and Rewitzky. We also exhibit a discrete duality for Urquhart’s relevant algebras and their frames.