Search

While the emergence of large administrative claims data provides opportunities for research, their use remains limited by the lack of clinical annotations relevant to disease outcomes, such as recurrence in breast cancer (BC). Several challenges arise from the annotation of such endpoints in administrative claims, including the need to infer both the occurrence and the date of the recurrence, the right-censoring of data, or the importance of time intervals between medical visits. Deep learning approaches have been successfully used to label temporal medical sequences, but no method is currently able to handle simultaneously right-censoring and visit temporality to detect survival events in medical sequences. We propose EDEN (Event DEtection Network), a time-aware Long-Short-Term-Memory network for survival analyses, and its custom loss function. Our method outperforms several state-of-the-art approaches on real-world BC datasets. EDEN constitutes a powerful tool to annotate disease recurrence from administrative claims, thus paving the way for the massive use of such data in BC research., Comment: Extended Abstract presented at Machine Learning for Health (ML4H) symposium 2022, November 28th, 2022, New Orleans, United States & Virtual, http://www.ml4h.cc, 6 pages

Besides reproducing tabular data properties of standalone tables, synthetic relational databases also require modeling the relationships between related tables. In this paper, we propose the Row Conditional-Tabular Generative Adversarial Network (RC-TGAN), a novel generative adversarial network (GAN) model that extends the tabular GAN to support modeling and synthesizing relational databases. The RC-TGAN models relationship information between tables by incorporating conditional data of parent rows into the design of the child table's GAN. We further extend the RC-TGAN to model the influence that grandparent table rows may have on their grandchild rows, in order to prevent the loss of this connection when the rows of the parent table fail to transfer this relationship information. The experimental results, using eight real relational databases, show significant improvements in the quality of the synthesized relational databases when compared to the benchmark system, demonstrating the effectiveness of the RC-TGAN in preserving relationships between tables of the original database.

We study from the point of view of rational equivalence the enveloping algebras of Lie algebras of dimension 3 whose derived Lie subalgebra is of dimension 2, over an algebraically closed base field in arbitrary characteristics.

This handbook outlines all test methods developed under the Development and Execution of Comprehensive and Integrated Subterranean Intelligent Vehicle Evaluations (DECISIVE) project by the University of Massachusetts Lowell for evaluating small unmanned aerial systems (sUAS) performance in subterranean and constrained indoor environments, spanning communications, field readiness, interface, obstacle avoidance, navigation, mapping, autonomy, trust, and situation awareness. For sUAS deployment in subterranean and constrained indoor environments, this puts forth two assumptions about applicable sUAS to be evaluated using these test methods: (1) able to operate without access to GPS signal, and (2) width from prop top to prop tip does not exceed 91 cm (36 in) wide (i.e., can physically fit through a typical doorway, although successful navigation through is not guaranteed). All test methods are specified using a common format: Purpose, Summary of Test Method, Apparatus and Artifacts, Equipment, Metrics, Procedure, and Example Data. All test methods are designed to be run in real-world environments (e.g., MOUT sites) or using fabricated apparatuses (e.g., test bays built from wood, or contained inside of one or more shipping containers)., Comment: Approved for public release: PAO #PR2022_47058

Striatal cholinergic interneurons (CINs) use acetylcholine (ACh) and glutamate (Glut) to regulate the striatal network since they express vesicular transporters for ACh (VAChT) and Glut (VGLUT3). However, whether ACh and Glut are released simultaneously and/or independently from cholinergic varicosities is an open question. The answer to that question requires the multichannel detection of vesicular transporters at the level of single synaptic vesicle (SV). Here, we used super-resolution STimulated Emission Depletion microscopy (STED) to characterize and quantify the distribution of VAChT and VGLUT3 in CINs SVs. Nearest-neighbor distances analysis between VAChT and VGLUT3-immunofluorescent spots revealed that 34% of CINs SVs contain both VAChT and VGLUT3. In addition, 40% of SVs expressed only VAChT while 26% of SVs contain only VGLUT3. These results suggest that SVs from CINs have the potential to store simultaneously or independently ACh and/or Glut. Overall, these morphological findings support the notion that CINs varicosities can signal with either ACh or Glut or both with an unexpected level of complexity.

Automated scoring for divergent thinking (DT) seeks to overcome a key obstacle to creativity measurement: the effort, cost, and reliability of scoring open-ended tests. For a common test of DT, the Alternate Uses Task (AUT), the primary automated approach casts the problem as a semantic distance between a prompt and the resulting idea in a text model. This work presents an alternative approach that greatly surpasses the performance of the best existing semantic distance approaches. Our system, "Ocsai," fine-tunes deep neural network-based large-language models (LLMs) on human-judged responses. Trained and evaluated against one of the largest collections of human-judged AUT responses, with 27 thousand responses collected from nine past studies, our fine-tuned large-language-models achieved up to r = 0.81 correlation with human raters, greatly surpassing current systems (r = 0.12-0.26). Further, learning transfers well to new test items and the approach is still robust with small numbers of training labels. We also compare prompt-based zero-shot and few-shot approaches, using GPT-3, ChatGPT, and GPT-4. This work also suggests a limit to the underlying assumptions of the semantic distance model, showing that a purely semantic approach that uses the stronger language representation of LLMs, while still improving on existing systems, does not achieve comparable improvements to our fine-tuned system. The increase in performance can support stronger applications and interventions in DT and opens the space of automated DT scoring to new areas for improving and understanding this branch of methods. [This paper was published in "Thinking Skills and Creativity" v49 Article 101356 2023.]

This work deals with shape optimization for contact mechanics. More specifically, the linear elasticity model is considered under the small deformations hypothesis, and the elastic body is assumed to be in contact (sliding or with Tresca friction) with a rigid foundation. The mathematical formulations studied are two regularized versions of the original variational inequality: the penalty formulation and the augmented Lagrangian formulation. In order to get the shape derivatives associated to those two non-differentiable formulations, we follow an approach based on directional derivatives \cite{chaudet2020shape,chaudet2021shape}. This allows us to develop a gradient-based topology optimization algorithm, built on these derivatives and a level-set representation of shapes. The algorithm also benefits from a mesh-cutting technique, which gives an explicit representation of the shape at each iteration, and enables us to apply the boundary conditions strongly on the contact zone. The different steps of the method are detailed. Then, to validate the approach, some numerical results on two-dimensional and three-dimensional benchmarks are presented.

This thesis deals with shape optimization for contact mechanics. More specifically, the linear elasticity model is considered under the small deformations hypothesis, and the elastic body is assumed to be in contact (sliding or with Tresca friction) with a rigid foundation. The mathematical formulations studied are two regularized versions of the original variational inequality: the penalty formulation and the augmented Lagrangian formulation. In order to get the shape derivatives associated to those two non-differentiable formulations, we suggest an approach based on directional derivatives. Especially, we derive sufficient conditions for the solution to be shape differentiable. This allows to develop a gradient-based topology optimization algorithm, built on these derivatives and a level-set representation of shapes. The algorithm also benefits from a mesh-cutting technique, which gives an explicit representation of the shape at each iteration, and enables to apply the boundary conditions strongly on the contact zone. The different steps of the method are detailed. Then, to validate the approach, some numerical results on two-dimensional and three-dimensional benchmarks are presented., Comment: PhD thesis, in French language

Business Process Simulation (BPS) is a common technique to estimate the impact of business process changes, e.g. what would be the cycle time of a process if the number of traces increases? The starting point of BPS is a business process model annotated with simulation parameters (a BPS model). Several studies have proposed methods to automatically discover BPS models from event logs -- extracted from enterprise information systems -- via process mining techniques. These approaches model the processing time of each activity based on the start and end timestamps recorded in the event log. In practice, however, it is common that the recorded start times do not precisely reflect the actual start of the activities. For example, a resource starts working on an activity, but its start time is not recorded until she/he interacts with the system. If not corrected, these situations induce waiting times in which the resource is considered to be free, while she/he is actually working. To address this limitation, this article proposes a technique to identify the waiting time previous to each activity instance in which the resource is actually working on them, and repair their start time so that they reflect the actual processing time. The idea of the proposed technique is that, as far as simulation is concerned, an activity instance may start once it is enabled and the corresponding resource is available. Accordingly, for each activity instance, the proposed technique estimates the activity enablement and the resource availability time based on the information available in the event log, and repairs the start time to include the non-recorded processing time. An empirical evaluation involving eight real-life event logs shows that the proposed approach leads to BPS models that closely reflect the temporal dynamics of the process., Comment: arXiv admin note: text overlap with arXiv:2206.14051

Business process simulation is a versatile technique to predict the impact of one or more changes on the performance of a process. Mainstream approaches in this space suffer from various limitations, some stemming from the fact that they treat resources as undifferentiated entities grouped into resource pools. These approaches assume that all resources in a pool have the same performance and share the same availability calendars. Previous studies have acknowledged these assumptions, without quantifying their impact on simulation model accuracy. This paper addresses this gap in the context of simulation models automatically discovered from event logs. The paper proposes a simulation approach and a method for discovering simulation models, wherein each resource is treated as an individual entity, with its own performance and availability calendar. An evaluation shows that simulation models with differentiated resources more closely replicate the distributions of cycle times and the work rhythm in a process than models with undifferentiated resources., Comment: Published in the Proceedings of the International Conference on Business Process Management (BPM'2022)

The Sample Adaptive Offset (SAO) filter has been introduced in HEVC to reduce general coding and banding artefacts in the reconstructed pictures, in complement to the De-Blocking Filter (DBF) which reduces artifacts at block boundaries specifically. The new video compression standard Versatile Video Coding (VVC) reduces the BD-rate by about 36% at the same reconstruction quality compared to HEVC. It implements an additional new in-loop Adaptive Loop Filter (ALF) on top of the DBF and the SAO filter, the latter remaining unchanged compared to HEVC. However, the relative performance of SAO in VVC has been lowered significantly. In this paper, it is proposed to revisit the SAO filter using Neural Networks (NN). The general principles of the SAO are kept, but the a-priori classification of SAO is replaced with a set of neural networks that determine which reconstructed samples should be corrected and in which proportion. Similarly to the original SAO, some parameters are determined at the encoder side and encoded per CTU. The average BD-rate gain of the proposed SAO improves VVC by at least 2.3% in Random Access while the overall complexity is kept relatively small compared to other NN-based methods.

We show that if the edges or vertices of an undirected graph $G$ can be covered by $k$ shortest paths, then the pathwidth of $G$ is upper-bounded by a single-exponential function of $k$. As a corollary, we prove that the problem Isometric Path Cover with Terminals (which, given a graph $G$ and a set of $k$ pairs of vertices called terminals, asks whether $G$ can be covered by $k$ shortest paths, each joining a pair of terminals) is FPT with respect to the number of terminals. The same holds for the similar problem Strong Geodetic Set with Terminals (which, given a graph $G$ and a set of $k$ terminals, asks whether there exist $\binom{k}{2}$ shortest paths covering $G$, each joining a distinct pair of terminals). Moreover, this implies that the related problems Isometric Path Cover and Strong Geodetic Set (defined similarly but where the set of terminals is not part of the input) are in XP with respect to parameter $k$.

Business Process Simulation (BPS) is a common approach to estimate the impact of changes to a business process on its performance measures. For example, it allows us to estimate what would be the cycle time of a process if we automated one of its activities, or if some resources become unavailable. The starting point of BPS is a business process model annotated with simulation parameters (a BPS model). In traditional approaches, BPS models are manually designed by modeling specialists. This approach is time-consuming and error-prone. To address this shortcoming, several studies have proposed methods to automatically discover BPS models from event logs via process mining techniques. However, current techniques in this space discover BPS models that only capture waiting times caused by resource contention or resource unavailability. Oftentimes, a considerable portion of the waiting time in a business process corresponds to extraneous delays, e.g., a resource waits for the customer to return a phone call. This article proposes a method that discovers extraneous delays from event logs of business process executions. The proposed approach computes, for each pair of causally consecutive activity instances in the event log, the time when the target activity instance should theoretically have started, given the availability of the relevant resource. Based on the difference between the theoretical and the actual start times, the approach estimates the distribution of extraneous delays, and it enhances the BPS model with timer events to capture these delays. An empirical evaluation involving synthetic and real-life logs shows that the approach produces BPS models that better reflect the temporal dynamics of the process, relative to BPS models that do not capture extraneous delays., Comment: Extended version of the ICPM 2022 publication (see v1)

Process mining techniques enable analysts to identify and assess process improvement opportunities based on event logs. A common roadblock to process mining is that event logs may contain private information that cannot be used for analysis without consent. An approach to overcome this roadblock is to anonymize the event log so that no individual represented in the original log can be singled out based on the anonymized one. Differential privacy is an anonymization approach that provides this guarantee. A differentially private event log anonymization technique seeks to produce an anonymized log that is as similar as possible to the original one (high utility) while providing a required privacy guarantee. Existing event log anonymization techniques operate by injecting noise into the traces in the log (e.g., duplicating, perturbing, or filtering out some traces). Recent work on differential privacy has shown that a better privacy-utility tradeoff can be achieved by applying subsampling prior to noise injection. In other words, subsampling amplifies privacy. This paper proposes an event log anonymization approach called Libra that exploits this observation. Libra extracts multiple samples of traces from a log, independently injects noise, retains statistically relevant traces from each sample, and composes the samples to produce a differentially private log. An empirical evaluation shows that the proposed approach leads to a considerably higher utility for equivalent privacy guarantees relative to existing baselines.

Prescriptive process monitoring approaches leverage historical data to prescribe runtime interventions that will likely prevent negative case outcomes or improve a process's performance. A centerpiece of a prescriptive process monitoring method is its intervention policy: a decision function determining if and when to trigger an intervention on an ongoing case. Previous proposals in this field rely on intervention policies that consider only the current state of a given case. These approaches do not consider the tradeoff between triggering an intervention in the current state, given the level of uncertainty of the underlying predictive models, versus delaying the intervention to a later state. Moreover, they assume that a resource is always available to perform an intervention (infinite capacity). This paper addresses these gaps by introducing a prescriptive process monitoring method that filters and ranks ongoing cases based on prediction scores, prediction uncertainty, and causal effect of the intervention, and triggers interventions to maximize a gain function, considering the available resources. The proposal is evaluated using a real-life event log. The results show that the proposed method outperforms existing baselines regarding total gain., Comment: BPM 2022 (20th International Conference on Business Process Management)

Planning the defossilization of energy systems by facilitating high penetration of renewables and maintaining access to abundant and affordable primary energy resources is a nontrivial multi-objective problem. However, so far, most long-term policies to decrease the carbon footprint of our societies consider the cost of the system as the leading indicator in the energy system models. This paper is the first to develop a novel approach by adding the energy return on investment (EROI) to a whole energy system optimization model. We built the database with all EROI technologies and resources considered. In addition, moving away from fossil-based to carbon-neutral energy systems raises the issue of the uncertainty of low-carbon technologies and resource data. Thus, we conducted a global sensitivity analysis to identify the main parameters driving the variations in the EROI of the system. This novel approach can be applied to any energy system, and we use a real-world case study to illustrate the model: the 2035 Belgian energy system for several greenhouse gas emissions targets. The main results are threefold: (i) the EROI of the system decreases from 8.9 to 3.9 when greenhouse gas emissions are reduced by 5; (ii) the renewable fuels - mainly imported renewable gas - represent the largest share of the system primary energy mix; (iii) in the sensitivity analysis, the renewable fuels drive 67% of the variation of the EROI of the system for low greenhouse gas emissions scenarios. The decrease in the EROI raises questions about meeting the climate targets without adverse socio-economic impact. Most countries rely massively on fossil fuels, and they could encounter an EROI decline when shifting to carbon neutrality. Thus, this study demonstrates the importance of considering other criteria, such as EROI, in energy system models., Comment: Revised version of the paper submitted to Biophysical Economics and Sustainability for a second round of review

Decentralized Finance (DeFi) is a nascent set of financial services, using tokens, smart contracts, and blockchain technology as financial instruments. We investigate four possible drivers of DeFi returns: exposure to cryptocurrency market, the network effect, the investor's attention, and the valuation ratio. As DeFi tokens are distinct from classical cryptocurrencies, we design a new dedicated market index, denoted DeFiX. First, we show that DeFi tokens returns are driven by the investor's attention on technical terms such as "decentralized finance" or "DeFi", and are exposed to their own network variables and cryptocurrency market. We construct a valuation ratio for the DeFi market by dividing the Total Value Locked (TVL) by the Market Capitalization (MC). Our findings do not support the TVL/MC predictive power assumption. Overall, our empirical study shows that the impact of the cryptocurrency market on DeFi returns is stronger than any other considered driver and provides superior explanatory power.

Machine learning models often fail to generalize well under distributional shifts. Understanding and overcoming these failures have led to a research field of Out-of-Distribution (OOD) generalization. Despite being extensively studied for static computer vision tasks, OOD generalization has been underexplored for time series tasks. To shine light on this gap, we present WOODS: eight challenging open-source time series benchmarks covering a diverse range of data modalities, such as videos, brain recordings, and sensor signals. We revise the existing OOD generalization algorithms for time series tasks and evaluate them using our systematic framework. Our experiments show a large room for improvement for empirical risk minimization and OOD generalization algorithms on our datasets, thus underscoring the new challenges posed by time series tasks. Code and documentation are available at https://woods-benchmarks.github.io ., Comment: 47 pages, 21 figures

Observatory end-to-end science operations is the overall process starting with a scientific question, represented by a proposal requesting observing time, and ending with the analysis of observation data addressing that question, and including all the intermediate steps needed to plan, schedule, obtain, and process these observations. Increasingly complex observing facilities demand a highly efficient science operations approach and at the same time be user friendly to the astronomical user community and enable the highest possible scientific return. Therefore, this process is supported by a collection of tools. In this paper, we describe the overall end-to-end process and its implementation for the three upcoming extremely large telescopes (ELTs), ESO's ELT, the Thirty Meter Telescope (TMT), and the Giant Magellan Telescope (GMT)., Comment: 38 pages, 20 figures. Revised submission to the SPIE Journal of Astronomical Telescopes, Instruments, and Systems

Many runners use watches incorporating Global Positioning System technology to track their workouts. These devices can be valuable training aids, but they have limitations. For several reasons including variations in satellite position, environmental factors, and design decisions made by the manufacturer, GPS-enabled watches can produce position measurement errors. These can result in incorrect estimations of total distance covered as well as running pace. This study examined the accuracy of three Garmin running watches of different technological generations using repetitive trials, over several years, by the same runner over the same route. The older watches, a Forerunner 205 and a Forerunner 220, showed similar accuracy when traversing the route. The newer generation watch, a Forerunner 45S, was found to be significantly less accurate in terms of both the trueness and precision of its distance measurements. This may indicate that Garmin, in competition with other manufacturers of similar devices, has chosen in recent years to prioritize device miniaturization and battery life over accuracy.