-
Sit Back and Relax: Learning to Drive Incrementally in All Weather Conditions
Authors:
Stefan Leitner,
M. Jehanzeb Mirza,
Wei Lin,
Jakub Micorek,
Marc Masana,
Mateusz Kozinski,
Horst Possegger,
Horst Bischof
Abstract:
In autonomous driving scenarios, current object detection models show strong performance when tested in clear weather. However, their performance deteriorates significantly when tested in degrading weather conditions. In addition, even when adapted to perform robustly in a sequence of different weather conditions, they are often unable to perform well in all of them and suffer from catastrophic fo…
▽ More
In autonomous driving scenarios, current object detection models show strong performance when tested in clear weather. However, their performance deteriorates significantly when tested in degrading weather conditions. In addition, even when adapted to perform robustly in a sequence of different weather conditions, they are often unable to perform well in all of them and suffer from catastrophic forgetting. To efficiently mitigate forgetting, we propose Domain-Incremental Learning through Activation Matching (DILAM), which employs unsupervised feature alignment to adapt only the affine parameters of a clear weather pre-trained network to different weather conditions. We propose to store these affine parameters as a memory bank for each weather condition and plug-in their weather-specific parameters during driving (i.e. test time) when the respective weather conditions are encountered. Our memory bank is extremely lightweight, since affine parameters account for less than 2% of a typical object detector. Furthermore, contrary to previous domain-incremental learning approaches, we do not require the weather label when testing and propose to automatically infer the weather condition by a majority voting linear classifier.
△ Less
Submitted 30 May, 2023;
originally announced May 2023.
-
Building resilient organizations: The roles of top-down vs. bottom-up organizing
Authors:
Stephan Leitner
Abstract:
Organizations face numerous challenges posed by unexpected events such as energy price hikes, pandemic disruptions, terrorist attacks, and natural disasters, and the factors that contribute to organizational success in dealing with such disruptions often remain unclear. This paper analyzes the roles of top-down and bottom-up organizational structures in promoting organizational resilience. To do s…
▽ More
Organizations face numerous challenges posed by unexpected events such as energy price hikes, pandemic disruptions, terrorist attacks, and natural disasters, and the factors that contribute to organizational success in dealing with such disruptions often remain unclear. This paper analyzes the roles of top-down and bottom-up organizational structures in promoting organizational resilience. To do so, an agent-based model of stylized organizations is introduced that features learning, adaptation, different modes of organizing, and environmental disruptions. The results indicate that bottom-up designed organizations tend to have a higher ability to absorb the effects of environmental disruptions, and situations are identified in which either top-down or bottom-up designed organizations have an advantage in recovering from shocks.
△ Less
Submitted 8 January, 2024; v1 submitted 12 May, 2023;
originally announced May 2023.
-
Designing organizations for bottom-up task allocation: The role of incentives
Authors:
Stephan Leitner
Abstract:
In recent years, various decentralized organizational forms have emerged, posing a challenge for organizational design. Some design elements, such as task allocation, become emergent properties that cannot be fully controlled from the top down. The central question that arises in this context is: How can bottom-up task allocation be guided towards an effective organizational structure? To address…
▽ More
In recent years, various decentralized organizational forms have emerged, posing a challenge for organizational design. Some design elements, such as task allocation, become emergent properties that cannot be fully controlled from the top down. The central question that arises in this context is: How can bottom-up task allocation be guided towards an effective organizational structure? To address this question, this paper presents a novel agent-based model of an organization that features bottom-up task allocation that can be motivated by either long-term or short-term orientation on the agents' side. The model also includes an incentive mechanism to guide the bottom-up task allocation process and create incentives that range from altruistic to individualistic. Our analysis shows that when bottom-up task allocation is driven by short-term orientation and aligned with the incentive mechanisms, it leads to improved organizational performance that surpasses that of traditionally designed organizations. Additionally, we find that the presence of altruistic incentive mechanisms within the organization reduces the importance of mirroring in task allocation.
△ Less
Submitted 1 January, 2023;
originally announced January 2023.
-
Collaborative search and autonomous task allocation in organizations of learning agents
Authors:
Stephan Leitner
Abstract:
This paper introduces a model of multi-unit organizations with either static structures, i.e., they are designed top-down following classical approaches to organizational design, or dynamic structures, i.e., the structures emerge over time from micro-level decisions. In the latter case, the units are capable of learning about the technical interdependencies of the task they face, and they use thei…
▽ More
This paper introduces a model of multi-unit organizations with either static structures, i.e., they are designed top-down following classical approaches to organizational design, or dynamic structures, i.e., the structures emerge over time from micro-level decisions. In the latter case, the units are capable of learning about the technical interdependencies of the task they face, and they use their knowledge by adapting the task allocation from time to time. In both static and dynamic organizations, searching for actions to increase the performance can either be carried out individually or collaboratively. The results indicate that (i) collaborative search processes can help overcome the adverse effects of inefficient task allocations as long as there is an internal fit with other organizational design elements, and (ii) for dynamic organizations, the emergent task allocation does not necessarily mirror the technical interdependencies of the task the organizations face, even though the same (or even higher) performances are achieved.
△ Less
Submitted 9 September, 2022; v1 submitted 5 June, 2022;
originally announced June 2022.
-
Balancing Consumer and Business Value of Recommender Systems: A Simulation-based Analysis
Authors:
Nada Ghanem,
Stephan Leitner,
Dietmar Jannach
Abstract:
Automated recommendations can nowadays be found on many e-commerce platforms, and such recommendations can create substantial value for consumers and providers. Often, however, not all recommendable items have the same profit margin, and providers might thus be tempted to promote items that maximize their profit. In the short run, consumers might accept non-optimal recommendations, but they may lo…
▽ More
Automated recommendations can nowadays be found on many e-commerce platforms, and such recommendations can create substantial value for consumers and providers. Often, however, not all recommendable items have the same profit margin, and providers might thus be tempted to promote items that maximize their profit. In the short run, consumers might accept non-optimal recommendations, but they may lose their trust in the long run. Ultimately, this leads to the problem of designing balanced recommendation strategies, which consider both consumer and provider value and lead to sustained business success. This work proposes a simulation framework based on agent-based modeling designed to help providers explore longitudinal dynamics of different recommendation strategies. In our model, consumer agents receive recommendations from providers, and the perceived quality of the recommendations influences the consumers' trust over time. We design several recommendation strategies which either give more weight on provider profit or on consumer utility. Our simulations show that a hybrid strategy that puts more weight on consumer utility but without ignoring profitability considerations leads to the highest cumulative profit in the long run. This hybrid strategy results in a profit increase of about 20 % compared to pure consumer or profit oriented strategies. We also find that social media can reinforce the observed phenomena. In case when consumers heavily rely on social media, the cumulative profit of the best strategy further increases. To ensure reproducibility and foster future research, we publicly share our flexible simulation framework.
△ Less
Submitted 18 August, 2022; v1 submitted 10 March, 2022;
originally announced March 2022.
-
Understanding Longitudinal Dynamics of Recommender Systems with Agent-Based Modeling and Simulation
Authors:
Gediminas Adomavicius,
Dietmar Jannach,
Stephan Leitner,
**g**g Zhang
Abstract:
Today's research in recommender systems is largely based on experimental designs that are static in a sense that they do not consider potential longitudinal effects of providing recommendations to users. In reality, however, various important and interesting phenomena only emerge or become visible over time, e.g., when a recommender system continuously reinforces the popularity of already successf…
▽ More
Today's research in recommender systems is largely based on experimental designs that are static in a sense that they do not consider potential longitudinal effects of providing recommendations to users. In reality, however, various important and interesting phenomena only emerge or become visible over time, e.g., when a recommender system continuously reinforces the popularity of already successful artists on a music streaming site or when recommendations that aim at profit maximization lead to a loss of consumer trust in the long run. In this paper, we discuss how Agent-Based Modeling and Simulation (ABM) techniques can be used to study such important longitudinal dynamics of recommender systems. To that purpose, we provide an overview of the ABM principles, outline a simulation framework for recommender systems based on the literature, and discuss various practical research questions that can be addressed with such an ABM-based simulation framework.
△ Less
Submitted 25 August, 2021;
originally announced August 2021.
-
On the Role of Incentives in Evolutionary Approaches to Organizational Design
Authors:
Stephan Leitner
Abstract:
This paper introduces a model of a stylized organization that is comprised of several departments that autonomously allocate tasks. To do so, the departments either take short-sighted decisions that immediately maximize their utility or take long-sighted decisions that aim at minimizing the interdependencies between tasks. The organization guides the departments' behavior by either an individualis…
▽ More
This paper introduces a model of a stylized organization that is comprised of several departments that autonomously allocate tasks. To do so, the departments either take short-sighted decisions that immediately maximize their utility or take long-sighted decisions that aim at minimizing the interdependencies between tasks. The organization guides the departments' behavior by either an individualistic, a balanced, or an altruistic linear incentive scheme. Even if tasks are perfectly decomposable, altruistic incentive schemes are preferred over individualistic incentive schemes since they substantially increase the organization's performance. Interestingly, if altruistic incentive schemes are effective, short-sighted decisions appear favorable since they do not only increase performance in the short run but also result in significantly higher performances in the long run.
△ Less
Submitted 11 May, 2021; v1 submitted 10 May, 2021;
originally announced May 2021.
-
An Agent-Based Model of Delegation Relationships With Hidden-Action: On the Effects of Heterogeneous Memory on Performance
Authors:
Patrick Reinwald,
Stephan Leitner,
Friederike Wall
Abstract:
We introduce an agent-based model of delegation relationships between a principal and an agent, which is based on the standard-hidden action model introduced by Holmström and, by doing so, provide a model which can be used to further explore theoretical topics in managerial economics, such as the efficiency of incentive mechanisms. We employ the concept of agentization, i.e., we systematically tra…
▽ More
We introduce an agent-based model of delegation relationships between a principal and an agent, which is based on the standard-hidden action model introduced by Holmström and, by doing so, provide a model which can be used to further explore theoretical topics in managerial economics, such as the efficiency of incentive mechanisms. We employ the concept of agentization, i.e., we systematically transform the standard hidden-action model into an agent-based model. Our modeling approach allows for a relaxation of some of the rather "heroic" assumptions included in the standard hidden-action model, whereby we particularly focus on assumptions related to the (i) availability of information about the environment and the (ii) principal's and agent's cognitive capabilities (with a particular focus on their learning capabilities and their memory). Our analysis focuses on how close and how fast the incentive scheme, which endogenously emerges from the agent-based model, converges to the solution proposed by the standard hidden-action model. Also, we investigate whether a stable solution can emerge from the agent-based model variant. The results show that in stable environments the emergent result can nearly reach the solution proposed by the standard hidden-action model. Surprisingly, the results indicate that turbulence in the environment leads to stability in earlier time periods.
△ Less
Submitted 28 September, 2020; v1 submitted 9 September, 2020;
originally announced September 2020.
-
On the Effectiveness of Minisum Approval Voting in an Open Strategy Setting: An Agent-Based Approach
Authors:
Joop van de Heijning,
Stephan Leitner,
Alexandra Rausch
Abstract:
This work researches the impact of including a wider range of participants in the strategy-making process on the performance of organizations which operate in either moderately or highly complex environments. Agent-based simulation demonstrates that the increased number of ideas generated from larger and diverse crowds and subsequent preference aggregation lead to rapid discovery of higher peaks i…
▽ More
This work researches the impact of including a wider range of participants in the strategy-making process on the performance of organizations which operate in either moderately or highly complex environments. Agent-based simulation demonstrates that the increased number of ideas generated from larger and diverse crowds and subsequent preference aggregation lead to rapid discovery of higher peaks in the organization's performance landscape. However, this is not the case when the expansion in the number of participants is small. The results confirm the most frequently mentioned benefit in the Open Strategy literature: the discovery of better performing strategies.
△ Less
Submitted 25 September, 2020; v1 submitted 7 September, 2020;
originally announced September 2020.
-
Understanding the dynamics emerging from infodemics: A call to action for interdisciplinary research
Authors:
Stephan Leitner,
Bartosz Gula,
Dietmar Jannach,
Ulrike Krieg-Holz,
Friederike Wall
Abstract:
Research on infodemics, i.e., the rapid spread of (mis)information related to a hazardous event, such as the COVID-19 pandemic, requires the integration of a multiplicity of scientific disciplines. The dynamics emerging from infodemics have the potential to generate complex behavioral patterns. In order to react appropriately, it is of ultimate importance for the fields of Business and Economics t…
▽ More
Research on infodemics, i.e., the rapid spread of (mis)information related to a hazardous event, such as the COVID-19 pandemic, requires the integration of a multiplicity of scientific disciplines. The dynamics emerging from infodemics have the potential to generate complex behavioral patterns. In order to react appropriately, it is of ultimate importance for the fields of Business and Economics to understand the dynamics emerging from it. In the short run, dynamics might lead to an adaptation in household spending or to a shift in buying behavior towards online providers. In the long run, changes in investments, consumer behavior, and markets are to be expected. We argue that the dynamics emerge from complex interactions among multiple factors, such as information and misinformation accessible for individuals and the formation and revision of beliefs. (Mis)information accessible to individuals is, amongst others, affected by algorithms specifically designed to provide personalized information, while automated fact-checking algorithms can help reduce the amount of circulating misinformation. The formation and revision of individual (and probably false) beliefs and individual fact-checking and interpretation of information are heavily affected by linguistic patterns inherent to information during pandemics and infodemics and further factors, such as affect, intuition and motives. We argue that, in order to get a deep(er) understanding of the dynamics emerging from infodemics, the fields of Business and Economics should integrate the perspectives of Computer Science and Information Systems, (Computational) Linguistics, and Cognitive Science into the wider context of economic systems (e.g., organizations, markets or industries) and propose a way to do so.
△ Less
Submitted 1 October, 2020; v1 submitted 23 July, 2020;
originally announced July 2020.
-
On the dynamics emerging from pandemics and infodemics
Authors:
Stephan Leitner
Abstract:
This position paper discusses emerging behavioral, social, and economic dynamics related to the COVID-19 pandemic and puts particular emphasis on two emerging issues: First, delayed effects (or second strikes) of pandemics caused by dread risk effects are discussed whereby two factors which might influence the existence of such effects are identified, namely the accessibility of (mis-)information…
▽ More
This position paper discusses emerging behavioral, social, and economic dynamics related to the COVID-19 pandemic and puts particular emphasis on two emerging issues: First, delayed effects (or second strikes) of pandemics caused by dread risk effects are discussed whereby two factors which might influence the existence of such effects are identified, namely the accessibility of (mis-)information and the effects of policy decisions on adaptive behavior. Second, the issue of individual preparedness to hazardous events is discussed. As events such as the COVID-19 pandemic unfolds complex behavioral patterns which are hard to predict, sophisticated models which account for behavioral, social, and economic dynamics are required to assess the effectivity and efficiency of decision-making.
△ Less
Submitted 19 April, 2020;
originally announced April 2020.
