-
Patients With Personality: Realistic Patient Simulation through Controlled Diversity and Selective Disclosure
Authors:
Moritz Schlager,
Friederike Jungmann,
Samuel Schmidgall,
Philipp Raffler,
Franziska Hartl,
Eva Wende,
Paula Roßmüller,
Conrad Ketzer,
Avinatan Hassidim,
Dale R. Webster,
Yossi Matias,
Yun Liu,
Daniel Rueckert,
Mike Schaekermann,
Paul Hager
Abstract:
Simulating realistic patient interactions is a key requirement to testing clinical applications of LLMs at scale without time-consuming and expensive user studies. However, existing approaches often lack realism and controllability, often oversharing information unprompted, and failing to capture the wide variability of patient behavior. Here, we introduce PatientsWithPersonality (PWP), a patient…
▽ More
Simulating realistic patient interactions is a key requirement to testing clinical applications of LLMs at scale without time-consuming and expensive user studies. However, existing approaches often lack realism and controllability, often oversharing information unprompted, and failing to capture the wide variability of patient behavior. Here, we introduce PatientsWithPersonality (PWP), a patient simulation framework that generates realistic yet diverse virtual patient responses through explicit personality parametrization over a latent patient state. Grounded in HEXACO, a six-dimensional personality space used to quantify and parameterize human behavioral traits, our approach enables fine-grained control over conversational style, cooperativeness, and information disclosure within a unified framework. In a clinician evaluation, PWP is judged nearly as realistic as recorded human actors and clearly ahead of prior simulators, while being flagged as "too informative" far less often. Conditioning on HEXACO axes yields personas whose configured traits are recoverable by both clinicians and an autorater, span a substantially wider behavioral footprint than the closest baseline, and prevent oversharing. Altogether, our framework paves the way for more accurate and informative LLM benchmarking through our realistic and steerable patient simulator.
△ Less
Submitted 11 August, 2026; v1 submitted 13 May, 2026;
originally announced June 2026.
-
Secure Decentralized Online Gaming with Lending Functionalities
Authors:
Katharina Alefs,
Florian Hartl,
Luke Newman,
Banu Ozdeveci,
Wisnu Uriawan
Abstract:
We present a decentralized online gaming platform implemented as a Decentralized Application (DApp) on the Ethereum blockchain. The gaming platform enables secure gaming, where the account balances and the stakes of the players are secured by a smart contract. Moreover, the fair enforcement of the game rules and the deposit of the winnings of the players and the gaming platform into their accounts…
▽ More
We present a decentralized online gaming platform implemented as a Decentralized Application (DApp) on the Ethereum blockchain. The gaming platform enables secure gaming, where the account balances and the stakes of the players are secured by a smart contract. Moreover, the fair enforcement of the game rules and the deposit of the winnings of the players and the gaming platform into their accounts are guaranteed by the smart contract. The gaming platform proposes lending functionalities that allow players to securely borrow tokens from the gaming platform in order to participate in the games.
△ Less
Submitted 4 May, 2022;
originally announced May 2022.
-
Visual Analysis of Multi-Parameter Distributions across Ensembles
Authors:
Alexander Kumpf,
Josef Stumpfegger,
Patrick Fabian Härtl,
Rüdiger Westermann
Abstract:
For an ensemble of data points in a multi-parameter space, we present a visual analytics technique to select a representative distribution of parameter values, and analyse how representative this distribution is in all ensemble members. A multi-parameter cluster in a representative ensemble member is visualized via a parallel coordinates plot, to provide initial distributions and let domain expert…
▽ More
For an ensemble of data points in a multi-parameter space, we present a visual analytics technique to select a representative distribution of parameter values, and analyse how representative this distribution is in all ensemble members. A multi-parameter cluster in a representative ensemble member is visualized via a parallel coordinates plot, to provide initial distributions and let domain experts interactively select relevant parameters and value ranges. Since unions of value ranges select hyper-cubes in parameter space, data points in these unions are not necessarily contained in the cluster. By using a multi-parameter kD-tree to further refine the selected parameter ranges, in combination with a covariance analysis of refined sets of data points, a tight partition in multi-parameter space with reduced number of falsely selected points is obtained. To assess the representativeness of the selected multi-parameter distribution across the ensemble, a linked side-by-side view of per-member violin plots is provided. We propose modifications of violin plots to show multi-parameter distributions simultaneously, and investigate the visual design that effectively conveys (dis-)similarities in multi-parameter distributions. In a linked spatial view, users can analyse and compare the spatial distribution of selected points in different ensemble members via interval-based isosurface raycasting. In two real-world application cases we show how our approach is used to analyse the multi-parameter distributions across an ensemble of 3D fields.
△ Less
Submitted 30 July, 2020;
originally announced July 2020.
-
Building a Reproducible Machine Learning Pipeline
Authors:
Peter Sugimura,
Florian Hartl
Abstract:
Reproducibility of modeling is a problem that exists for any machine learning practitioner, whether in industry or academia. The consequences of an irreproducible model can include significant financial costs, lost time, and even loss of personal reputation (if results prove unable to be replicated). This paper will first discuss the problems we have encountered while building a variety of machine…
▽ More
Reproducibility of modeling is a problem that exists for any machine learning practitioner, whether in industry or academia. The consequences of an irreproducible model can include significant financial costs, lost time, and even loss of personal reputation (if results prove unable to be replicated). This paper will first discuss the problems we have encountered while building a variety of machine learning models, and subsequently describe the framework we built to tackle the problem of model reproducibility. The framework is comprised of four main components (data, feature, scoring, and evaluation layers), which are themselves comprised of well defined transformations. This enables us to not only exactly replicate a model, but also to reuse the transformations across different models. As a result, the platform has dramatically increased the speed of both offline and online experimentation while also ensuring model reproducibility.
△ Less
Submitted 9 October, 2018;
originally announced October 2018.
-
The Electric Two-echelon Vehicle Routing Problem
Authors:
Ulrich Breunig,
Roberto Baldacci,
Richard F. Hartl,
Thibaut Vidal
Abstract:
Two-echelon distribution systems are attractive from an economical standpoint and help to keep large vehicles out of city centers. Large trucks can be used to deliver goods to intermediate facilities in accessible locations, whereas smaller vehicles allow to reach the final customers. Due to their reduced size and emissions, companies consider using an electric fleet of terrestrian or aerial vehic…
▽ More
Two-echelon distribution systems are attractive from an economical standpoint and help to keep large vehicles out of city centers. Large trucks can be used to deliver goods to intermediate facilities in accessible locations, whereas smaller vehicles allow to reach the final customers. Due to their reduced size and emissions, companies consider using an electric fleet of terrestrian or aerial vehicles for last mile deliveries. Route planning in multi-tier logistics leads to notoriously difficult problems. This difficulty is accrued in the presence of an electric fleet, since each vehicle operates on a smaller range, and may require visits to charging stations. To study these challenges, we introduce the Electric Two-echelon Vehicle Routing Problem as a prototypical problem. We propose a large neighbourhood search metaheuristic as well as an exact mathematical programming algorithm, which uses decomposition techniques to enumerate promising first-level solutions, in conjunction with bounding functions and route enumeration for the second-level routes. These algorithms produce optimal or near-optimal solutions for the problem, and allow us to evaluate the impact of several defining features of optimized battery-powered distribution networks. We created representative E2EVRP benchmark instances to simulate realistic metropolitan areas. In particular, we observe that the detour miles due to recharging decrease proportionally to $1/ρ^x$ with $x \approx 5/4$ as a function of the charging stations density $ρ$; e.g., in a scenario where the density of charging stations is doubled, recharging detours are reduced by 58\%. Finally, we evaluate the trade-off between battery capacity and detour miles. This estimate is critical for strategic fleet-acquisition decisions, in a context where large batteries are generally more costly and less environment-friendly.
△ Less
Submitted 15 November, 2018; v1 submitted 9 March, 2018;
originally announced March 2018.
-
Collaborative vehicle routing: a survey
Authors:
Margaretha Gansterer,
Richard F. Hartl
Abstract:
In horizontal collaborations, carriers form coalitions in order to perform parts of their logistics operations jointly. By exchanging transportation requests among each other, they can operate more efficiently and in a more sustainable way. Collaborative vehicle routing has been extensively discussed in the literature. We identify three major streams of research: (i) centralized collaborative plan…
▽ More
In horizontal collaborations, carriers form coalitions in order to perform parts of their logistics operations jointly. By exchanging transportation requests among each other, they can operate more efficiently and in a more sustainable way. Collaborative vehicle routing has been extensively discussed in the literature. We identify three major streams of research: (i) centralized collaborative planning, (ii) decentralized planning without auctions, and (ii) auction-based decentralized planning. For each of them we give a structured overview on the state of knowledge and discuss future research directions.
△ Less
Submitted 13 June, 2017;
originally announced June 2017.
-
Capacity Planning for Cluster Tools in the Semiconductor Industry
Authors:
Martin Romauch,
Richard F. Hartl
Abstract:
This paper proposes a new model for Cluster-tools with two load locks. Cluster-tools are widely used to automate single wafer processing in semiconductor industry. The load locks are the entry points into the vacuum of the Cluster-tool's mainframe. Usually there are two of them available. Each lot being processed, is dedicated to a single load-lock. Therefore at most two different lots (with possi…
▽ More
This paper proposes a new model for Cluster-tools with two load locks. Cluster-tools are widely used to automate single wafer processing in semiconductor industry. The load locks are the entry points into the vacuum of the Cluster-tool's mainframe. Usually there are two of them available. Each lot being processed, is dedicated to a single load-lock. Therefore at most two different lots (with possibly different processing times and qualification) can be processed simultaneously. This restriction is one of the major potential bottlenecks.
Capacity planning is one of the possible applications for the proposed model and the paper demonstrates the integration into a more general framework that considers different tool types and different operational modes.
The paper also generalizes an earlier model that is limited to three processing chambers. The proposed modeling approach is based on makespan reductions by parallel processing. It turns out that the performance of the new approach is similar, when compared to the generalized model for three chambers, but the new approach outperforms the generalized model for four and more chambers.
△ Less
Submitted 3 May, 2016;
originally announced May 2016.
-
The lateral transhipment problem with a-priori routes, and a lot sizing application
Authors:
Martin Romauch,
Thibaut Vidal,
Richard F. Hartl
Abstract:
We propose exact solution approaches for a lateral transhipment problem which, given a pre-specified sequence of customers, seeks an optimal inventory redistribution plan considering travel costs and profits dependent on inventory levels. Trip-duration and vehicle-capacity constraints are also imposed. The same problem arises in some lot sizing applications, in the presence of setup costs and equi…
▽ More
We propose exact solution approaches for a lateral transhipment problem which, given a pre-specified sequence of customers, seeks an optimal inventory redistribution plan considering travel costs and profits dependent on inventory levels. Trip-duration and vehicle-capacity constraints are also imposed. The same problem arises in some lot sizing applications, in the presence of setup costs and equipment re-qualifications.
We introduce a pure dynamic programming approach and a branch-and-bound framework that combines dynamic programming with Lagrangian relaxation. Computational experiments are conducted to determine the most suitable solution approach for different instances, depending on their size, vehicle capacities and duration constraints. The branch-and-bound approach, in particular, solves problems with up to 50 delivery locations in less than ten seconds on a modern computer.
△ Less
Submitted 23 December, 2015;
originally announced December 2015.
-
A large neighbourhood based heuristic for two-echelon routing problems
Authors:
Ulrich Breunig,
Verena Schmid,
Richard F. Hartl,
Thibaut Vidal
Abstract:
In this paper, we address two optimisation problems arising in the context of city logistics and two-level transportation systems. The two-echelon vehicle routing problem and the two-echelon location routing problem seek to produce vehicle itineraries to deliver goods to customers, with transits through intermediate facilities. To efficiently solve these problems, we propose a hybrid metaheuristic…
▽ More
In this paper, we address two optimisation problems arising in the context of city logistics and two-level transportation systems. The two-echelon vehicle routing problem and the two-echelon location routing problem seek to produce vehicle itineraries to deliver goods to customers, with transits through intermediate facilities. To efficiently solve these problems, we propose a hybrid metaheuristic which combines enumerative local searches with destroy-and-repair principles, as well as some tailored operators to optimise the selections of intermediate facilities. We conduct extensive computational experiments to investigate the contribution of these operators to the search performance, and measure the performance of the method on both problem classes. The proposed algorithm finds the current best known solutions, or better ones, for 95% of the two-echelon vehicle routing problem benchmark instances. Overall, for both problems, it achieves high-quality solutions within short computing times. Finally, for future reference, we resolve inconsistencies between different versions of benchmark instances, document their differences, and provide them all online in a unified format.
△ Less
Submitted 4 October, 2016; v1 submitted 29 May, 2015;
originally announced May 2015.