BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//scienceofintelligence.de - ECPv6.16.3//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-ORIGINAL-URL:https://www.scienceofintelligence.de
X-WR-CALDESC:Events for scienceofintelligence.de
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Europe/Berlin
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20220327T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20221030T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20230326T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20231029T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20240331T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20241027T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20250330T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20251026T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20260329T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20261025T010000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20250116T100000
DTEND;TZID=Europe/Berlin:20250116T110000
DTSTAMP:20260615T153633
CREATED:20250106T095531Z
LAST-MODIFIED:20250107T153234Z
UID:22991-1737021600-1737025200@www.scienceofintelligence.de
SUMMARY:Anita Keshmirian (Forward College\, Berlin): "Many Minds\, Diverging Morals: Human Groups vs. AI in Moral Decision-Making"
DESCRIPTION:Moral judgments are inherently social\, shaped by interactions with others in everyday life. Despite this\, psychological research has rarely examined the impact of social interactions on these judgments. In our study\, we explored the role of group dynamics in moral decision-making by having small groups (4-5 participants) evaluate moral dilemmas first individually\, then collectively\, and finally individually a second time. Participants judged real-life and sacrificial moral dilemmas involving actions or inactions violating moral principles to benefit the greater good. Experiment 1 found that collective judgments were more utilitarian than individual judgments\, supporting the hypothesis that group deliberation temporarily reduces the emotional burden of violating moral norms. \nExperiment 2 measured participants’ state anxiety and moral judgments before\, during\, and after online interactions. Results again showed that collectives were more utilitarian\, reducing state anxiety during and after social interaction\, suggesting that stress reduction may explain the shift toward utilitarianism in group settings. We replicated this experiment using multi-agent large language models (LLMs) to test how artificial agents make moral decisions. Preliminary findings revealed that\, unlike humans\, groups of LLM agents were less utilitarian than individual agents. Analysis of the agents’ interactions showed a consistent pattern of virtue-signaling\, with LLMs emphasizing deontological reasoning (focusing on moral rules) rather than utilitarian principles. \nThis divergence from human behavior suggests that collective reasoning in AI systems is shaped by different dynamics\, likely due to how LLMs are trained to prioritize socially accepted norms. These results highlight important differences in moral decision-making between human and artificial intelligence\, offering new insights into the development of AI systems that more closely mirror human ethical reasoning\, particularly in complex\, real-world collective decision-making scenarios. \nImage created with DALL-E by Maria Ott
URL:https://www.scienceofintelligence.de/event/anita-keshmirian-many-minds-diverging-morals-human-groups-vs-ai-in-moral-decision-making/
LOCATION:SCIoI\, Marchstraße 23\, 10587 Berlin\, Room 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/webp:https://www.scienceofintelligence.de/wp-content/uploads/2025/01/TMT_Anita_Keshmirian-2-e1736256383948.webp
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20241128T100000
DTEND;TZID=Europe/Berlin:20241128T110000
DTSTAMP:20260615T153633
CREATED:20240911T084638Z
LAST-MODIFIED:20250603T124540Z
UID:22176-1732788000-1732791600@www.scienceofintelligence.de
SUMMARY:Hideki Kozima (Tohoku University)\, “Child-Robot Interactions for Therapeutic and Educational Research and Practices”
DESCRIPTION:Abstract:\nResearch in developmental robotics includes modeling human intelligence and the process of its emergence in robotic systems. A novel research paradigm in psychology is emerging in conjunction with such efforts regarding reproducing human-specific communication abilities in robots and observing how children interact with robots with various communication capabilities. I will discuss such research trends from a broader perspective\, and the potential to realize robots that afford children the opportunities to build social relationships is examined. We consider human communication abilities not as a set of interactive functionalities in individuals but as a set of social tools in which functionalities have emerged from the social interaction driven by the individual motivation to form relationships with others. We also explore some representative works on using robots to help establish such relationships in autistic and typically developing children. Finally\, we discuss the possibility of using robots for further research and practical support for child development centered on building social relationships. \n  \nThis talk will take place in person at SCIoI (room 2.057). \n  \nPhoto by Owen Beard on Unsplash.
URL:https://www.scienceofintelligence.de/event/hideki-kozima-tohuku-university/
LOCATION:SCIoI\, Marchstraße 23\, 10587 Berlin\, Room 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/09/xkozima-L-gray.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20241031T100000
DTEND;TZID=Europe/Berlin:20241031T110000
DTSTAMP:20260615T153633
CREATED:20241002T101830Z
LAST-MODIFIED:20250603T124704Z
UID:22386-1730368800-1730372400@www.scienceofintelligence.de
SUMMARY:POSTPONED: Anita Keshmirian (Forward College\, Berlin)\, “Many Minds\, Diverging Morals: Human Groups vs. AI in Moral Decision-Making”
DESCRIPTION:Abstract \n“Moral judgments are inherently social\, shaped by interactions with others in everyday life. Despite this\, psychological research has rarely examined the impact of social interactions on these judgments. In our study\, we explored the role of group dynamics in moral decision making by having small groups (4-5 participants) evaluate moral dilemmas first individually\, then collectively\, and finally individually a second time. Participants judged real-life and sacrificial moral dilemmas involving actions or inactions violating moral principles to benefit the greater good. Experiment 1 found that collective judgments were more utilitarian than individual judgments\, supporting the hypothesis that group deliberation temporarily reduces the emotional burden of violating moral norms. Experiment 2 measured participants’ state anxiety and moral judgments before\, during\, and after online interactions. Results again showed that collectives were more utilitarian\, reducing state anxiety during and after social interaction\, suggesting that stress reduction may explain the shift toward utilitarianism in group settings. \nWe replicated this experiment using multi-agent large language models (LLMs) to test how artificial agents make moral decisions. Preliminary findings revealed that\, unlike humans\, groups of LLM agents were less utilitarian than individual agents. Analysis of the agents’ interactions showed a consistent pattern of virtue-signaling\, with LLMs emphasizing deontological reasoning (focusing on moral rules) rather than utilitarian principles. This divergence from human behavior suggests that collective reasoning in AI systems is shaped by different dynamics\, likely due to how LLMs are trained to prioritize socially accepted norms. These results highlight important differences in moral decision-making between human and artificial intelligence\, offering new insights into the development of AI systems that more closely mirror human ethical reasoning\, particularly in complex\, real-world collective decision-making scenarios.” \nImage credit: ©SCIoI/ generated with DALL-E
URL:https://www.scienceofintelligence.de/event/anita-keshmirian-forward-college-berlin-many-minds-diverging-morals-human-groups-vs-ai-in-moral-decision-making/
LOCATION:Marchstraße 23\, 10587 Berlin\, Room 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/webp:https://www.scienceofintelligence.de/wp-content/uploads/2024/10/TMT_Image_creativity_artificial2-e1727864193669.webp
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240725T100000
DTEND;TZID=Europe/Berlin:20240725T110000
DTSTAMP:20260615T153633
CREATED:20240719T144755Z
LAST-MODIFIED:20250603T124813Z
UID:21025-1721901600-1721905200@www.scienceofintelligence.de
SUMMARY:Ralf M. Haefner (University of Rochester\, NY)\, “How We Move Our Eyes To Collect Information”
DESCRIPTION:Abstract\nCollecting new information about the outside world is a key aspect of brain function. In the context of vision\, we move our eyes multiple times per second to accumulate evidence about a scene. Prior studies have suggested that this process is goal-directed and close to optimal. Here\, we show that this process of seeking new information suffers from a confirmation bias similar to what has been observed in a wide range of other contexts. We present data from a new gaze-contingent task that allows us to both estimate a participant’s current belief\, and compare that to their subsequent eye-movements. We find that these eye-movements are biased in a confirmatory way. Finally\, we show that these empirical results can be parsimoniously explained under the assumption that the brain performs approximate\, not exact\, inference\, with computations being more approximate in decision-making compared to sensory areas. \nThis talk will take place in person at SCIoI. \n  \nImage created with DALL-E by Maria Ott.
URL:https://www.scienceofintelligence.de/event/ralf-m-haefner-university-of-rochester-ny-how-we-move-our-eyes-to-collect-information/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/webp:https://www.scienceofintelligence.de/wp-content/uploads/2024/07/Ralf-Haefner_1.webp
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240718T100000
DTEND;TZID=Europe/Berlin:20240718T113000
DTSTAMP:20260615T153633
CREATED:20240624T113208Z
LAST-MODIFIED:20250603T124835Z
UID:20867-1721296800-1721302200@www.scienceofintelligence.de
SUMMARY:Adrian Sieler (Science of Intelligence): “Building Anthropomorphic Soft Robotic Hands With Human-Like Manipulation Abilities”
DESCRIPTION:More info to follow. \n  \nThis talk will take place in person at SCIoI.
URL:https://www.scienceofintelligence.de/event/adrian-sieler-science-of-intelligence-building-anthropomorphic-soft-robotic-hands-with-human-like-manipulation-abilities/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/06/20201020-SCIOI-Adrian1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240704T100000
DTEND;TZID=Europe/Berlin:20240704T113000
DTSTAMP:20260615T153633
CREATED:20240506T082117Z
LAST-MODIFIED:20250603T124850Z
UID:19371-1720087200-1720092600@www.scienceofintelligence.de
SUMMARY:Caleb Weinreb (Harvard Medical School)\, “A Seconds-Long Timescale in Naturalistic Behavior Structures Neural Dynamics”
DESCRIPTION:A core task of animal cognition is to carve the world up into relevant contextual states – based on sensory input\, internal drives\, and awareness of one’s own recent behavior – and then hold these state assignments in working memory as guides for action and anchors for learning. By training animals to perform asks with well-defined contextual states\, researchers have homed in on prefrontal cortex (PFC) as a critical node for such contextual state inference. But these tasks are a poor approximation of real life; rather than engaging in a single well-defined task\, free animals define their own tasks and engage in them dynamically over time recognizing contexts that emerge naturally from their own interactions with task affordances. A core question in neuroethology is which specific “task states” emerge in a given experimental setting and how they structure neural dynamics\, including in PFC. We took advantage of motion sequencing (MoSeq) — which uses 3D pose tracking and machine learning to segment behavior into sub-second motifs or “syllables” – to understand how mPFC activity coevolves with behavior across multiple timescales during unconstrained social interaction and solitary exploration. We find mPFC activity correlates strongly with ongoing behavior\, and that these correlations are most parsimoniously explained through an underlying manifold of behavior states that evolve on a timescale of seconds. The behavior states influence not only which PFC neurons are active\, but also which variables are most strongly encoded. We also find that the composition of states is labile and propose that it emerges predictably from the number and salience of affordances in the animal’s environment. \n  \nThis talk will take place in person at SCIoI. \nPhoto by Pietro Jeng on Unsplash. \n 
URL:https://www.scienceofintelligence.de/event/caleb-weinreb-harvard-medical-school/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/05/pietro-jeng-n6B49lTx7NM-unsplash-1024x1024-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240613T100000
DTEND;TZID=Europe/Berlin:20240613T113000
DTSTAMP:20260615T153633
CREATED:20240530T134836Z
LAST-MODIFIED:20250603T124915Z
UID:20720-1718272800-1718278200@www.scienceofintelligence.de
SUMMARY:Aravind Battaje (Science of Intelligence)\, “A Study on Human and Robot Perception and the Architecture of Perceptual Information Processing”
DESCRIPTION:More details to follow. \n  \nThis talk will take place in person at SCIoI. \nPhoto by Markus Spiske on Unsplash. \n 
URL:https://www.scienceofintelligence.de/event/aravind-battaje-a-study-on-human-and-robot-perception-and-the-architecture-of-perceptual-information-processing/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/05/markus-spiske-Skf7HxARcoc-unsplash-scaled.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240530T100000
DTEND;TZID=Europe/Berlin:20240530T110000
DTSTAMP:20260615T153633
CREATED:20240424T083921Z
LAST-MODIFIED:20240813T103215Z
UID:19215-1717063200-1717066800@www.scienceofintelligence.de
SUMMARY:Verena Wagner (University of Konstanz)\, “On Pause: Suspending Judgment and Abstaining in Machine Learning”
DESCRIPTION:Abstract:\n\nMachine Learning (ML) systems typically yield definitive outputs\, even when the underlying probabilities do not justify a decision. This poses a significant challenge in medical applications\, where patients rely on individualized diagnoses\, treatments\, and prognoses. A recent advancement in ML research addresses this issue by introducing so-called “abstention models\,” which enable ML systems to provide neutral outputs. From the perspective of a philosopher who works on cognitive neutrality and the suspension of judgment in human agents\, this is an interesting field to explore. In this talk\, I will introduce my philosophical theory of cognitive neutrality\, which promotes various ways of suspending judgment. Against this backdrop\, I will explore different abstention models and look for similarities and differences between suspension of judgment in humans and abstention in ML systems. In particular\, I will examine whether the distinctions outlined in my cognitive neutrality framework also manifest in different models of abstention. \n  \n\n\n\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Markus Spiske on Unsplash. \n 
URL:https://www.scienceofintelligence.de/event/verena-wagner-university-of-konstanz-on-pause-suspending-judgment-and-abstaining-in-machine-learning/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/michael-dziedzic-aQYgUYwnCsM-uns-scaled.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240523T100000
DTEND;TZID=Europe/Berlin:20240523T110000
DTSTAMP:20260615T153633
CREATED:20240424T114021Z
LAST-MODIFIED:20250603T124944Z
UID:19227-1716458400-1716462000@www.scienceofintelligence.de
SUMMARY:Asieh Daneshi\, “The Effect of Group Size and Group Density on Behavioral Contagion in Humans”
DESCRIPTION:More details to follow. \n  \nThis talk will take place in person at SCIoI. \n 
URL:https://www.scienceofintelligence.de/event/asieh-daneshi-the-effect-of-group-size-and-group-density-on-behavioral-contagion-in-humans/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/IMG_6596-1536x1536-1-1024x1024-1-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240516T100000
DTEND;TZID=Europe/Berlin:20240516T110000
DTSTAMP:20260615T153633
CREATED:20240424T113512Z
LAST-MODIFIED:20250603T124959Z
UID:19222-1715853600-1715857200@www.scienceofintelligence.de
SUMMARY:Alicia Burns\, “Predator-Prey Interactions in the Open Ocean”
DESCRIPTION:More details to follow. \nThis talk will take place in person at SCIoI. \nPhoto by: Rodrigo Friscione Wyssmann. \n 
URL:https://www.scienceofintelligence.de/event/shepherding-behaviour-in-predator-prey-interactions-p33/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/png:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/Screenshot-2024-04-24-at-14.22.35.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240502T100000
DTEND;TZID=Europe/Berlin:20240502T110000
DTSTAMP:20260615T153633
CREATED:20240424T112306Z
LAST-MODIFIED:20240813T103302Z
UID:19218-1714644000-1714647600@www.scienceofintelligence.de
SUMMARY:Joshua B. Evans\, “Creating Multi-Level Skill Hierarchies in Reinforcement Learning”
DESCRIPTION:Abstract:\n\nWhat is a useful skill hierarchy for an autonomous agent? In this talk\, we will consider a possible answer based on a graphical representation of how the interaction between an agent and its environment may unfold. The proposed approach uses modularity maximisation as a central organising principle to expose the structure of the interaction graph at multiple levels of abstraction. The result is a collection of skills that operate at varying time scales\, organised into a hierarchy\, where skills that operate over longer time scales are composed of skills that operate over shorter time scales. The entire skill hierarchy is generated automatically\, with no human intervention\, including the skills themselves (their behaviour\, when they can be called\, and when they terminate) as well as the hierarchical dependency structure between them. In a wide range of environments\, this approach generates skill hierarchies that are intuitively appealing and that considerably improve the learning performance agents given access to them. \n  \n\n\n\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Bofu Shaw on Unsplash \n 
URL:https://www.scienceofintelligence.de/event/joshua-b-evans-creating-multi-level-skill-hierarchies-in-reinforcement-learning/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/bofu-shaw-yty30exygSI-unsplash-1536x1536-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240418T100000
DTEND;TZID=Europe/Berlin:20240418T113000
DTSTAMP:20260615T153633
CREATED:20231113T143920Z
LAST-MODIFIED:20250603T125021Z
UID:18049-1713434400-1713439800@www.scienceofintelligence.de
SUMMARY:Pavel Němec (Charles University)\, “Two Independent Origins of Complex Brains and Intelligent Behavior in Birds and Mammals”
DESCRIPTION:Abstract:\n\nOver the last 20 years\, it has been shown that birds and mammals are startlingly similar in their cognitive repertoire. Even the most intelligent taxa from each group – great apes and large corvids and parrots – match each other in most domains of cognition. This functional similarity is remarkable considering that birds and mammals shared a last common ancestor about 325 million years ago. Moreover\, avian brains are small and lack a cerebral cortex arranged in layers. My talk will focus on recent discoveries showing that birds and mammals independently evolved brains with dramatically increased neuron numbers in the telencephalon and cerebellum\, brain parts associated with higher cognition. This brain information processing capacity surge in birds and mammals is associated with the elaboration of at least partly non-homologous neural circuitry. Moreover\, similar functions are processed in different\, non-homological forebrain regions. Extreme neuron packing densities in birds partly explain why they have similar cognitive levels as mammals\, but volumetrically much smaller brains. Astoundingly\, phylogenetic analysis suggests that as few as four major changes in neuron-brain scaling in over 300 million years of evolution pave the way to intelligence in endothermic land vertebrates. \n\nPh. kindly provided by Pavel Němec. \n\n\n\n\nThis talk will take place in person at SCIoI. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-thursday-morning-talk-pavel-nemec-charles-university-two-independent-origins-of-complex-brains-and-intelligent-behavior-in-birds-and-mammals/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/png:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/b.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240411T100000
DTEND;TZID=Europe/Berlin:20240411T110000
DTSTAMP:20260615T153633
CREATED:20231113T093715Z
LAST-MODIFIED:20250603T125032Z
UID:17053-1712829600-1712833200@www.scienceofintelligence.de
SUMMARY:Hector Garcia De Marina (University of Granada)\, “Practical Challenges in Formation Control and Mobile Robot Swarms”
DESCRIPTION:Abstract: \nRobot swarms have the potential to assist us with simpler logistics in persistent missions involving vast scenarios. Robot swarms also promise added resilience to complete their objectives despite unforeseen difficulties. However\, current demonstrations of swarm technology in unstructured environments only count on single-digit individuals. That is farther from what one would expect from the huge scaling potential of a swarm. What are the bottlenecks then? \nIn this talk\, I will present some practical challenges that mobile robot swarms face in fundamental tasks\, e.g.\, the control of specific geometry parameters during a swarm deployment\, also known as formation control. As an application of higher-level tasks leveraging formation control\, we will see the coordination of robots while tracking paths and the source-seeking of scalar fields. \nI will also focus on onboard imperfections and how they are responsible for non-designed emergent behavior. Nevertheless\, I will show some hidden opportunities within the imperfections that could assist us with practical deployments. \n\n\nRelated articles (free links to Arxiv):\nManeuvering and robustness issues in undirected displacement-consensus-based formation control: https://arxiv.org/abs/2008.03544\nGuiding vector fields for the distributed motion coordination of mobile robots: https://arxiv.org/abs/2209.09478v4\nResilient source seeking with robot swarms: https://arxiv.org/abs/2309.02937\nBehavioral-based circular formation control for robot swarms: https://arxiv.org/abs/2309.09101\n\n\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Louis Reed on Unsplash. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-hector-garcia-de-marina/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/11/louis-reed-wSTCaQpiLtc-unsplash.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240404T100000
DTEND;TZID=Europe/Berlin:20240404T233000
DTSTAMP:20260615T153633
CREATED:20240321T143812Z
LAST-MODIFIED:20240813T103456Z
UID:18046-1712224800-1712273400@www.scienceofintelligence.de
SUMMARY:Jacek Wiland\,  "Assessing the Factual Knowledge Contained in Language Models During Lifelong Learning"
DESCRIPTION:More details to follow.\n\n\n\n\nThis talk will take place in person at SCIoI. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-jacek-wiland-assessing-the-factual-knowledge-contained-in-language-models-during-lifelong-learning/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/Jacek-Wiland-1024x1024-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240328T100000
DTEND;TZID=Europe/Berlin:20240328T113000
DTSTAMP:20260615T153633
CREATED:20240318T143711Z
LAST-MODIFIED:20250603T125108Z
UID:18043-1711620000-1711625400@www.scienceofintelligence.de
SUMMARY:Heiner Spieß (Science of Intelligence)\, “Tools to Study the Generality of Deep Neural Network Representations”
DESCRIPTION:More details to follow.\n\n\n\nThis talk will take place in person at SCIoI. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-heiner-spies-tools-to-study-the-generality-of-deep-neural-network-representations/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/Heiner2-1-768x768-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240321T100000
DTEND;TZID=Europe/Berlin:20240321T110000
DTSTAMP:20260615T153633
CREATED:20231113T093422Z
LAST-MODIFIED:20240813T103529Z
UID:17052-1711015200-1711018800@www.scienceofintelligence.de
SUMMARY:Fariba Karimi (Graz University of Technology)\, "Complexity Science for Societal Good"
DESCRIPTION:Abstract:\nSocial inequalities — structured and recurrent patterns of unequal distribution of wealth\, opportunities\, and rewards — are on the rise\, and quick-fix\, top-down approaches are failing. Structural inequality is one of the important manifestations of social inequalities in which institutions\, policies\, and societies create systems of privilege that are structural barriers to equality and inclusiveness. Structural inequalities emerge and evolve in complex multi-dimensional social networks. With the rise of artificial intelligence and algorithms in decision-making processes\, such inequalities are being reinforced and exacerbated in a non-linear\, complex manner that is difficult to comprehend and tackle. To address and mitigate such timely issues\, we need a complexity science approach and interdisciplinary teams more than ever.\n\n\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Vincentiu Solomon on Unsplash. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-fariba-karimi/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/11/vincentiu-solomon-IHnG5xfSZK0-unsplash.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240307T100000
DTEND;TZID=Europe/Berlin:20240307T230000
DTSTAMP:20260615T153633
CREATED:20240124T140430Z
LAST-MODIFIED:20250603T125133Z
UID:18023-1709805600-1709852400@www.scienceofintelligence.de
SUMMARY:Christian Poth (Bielefeld University)\, “Task-Driven Phasic Alertness: How Being Ready for Action Relies on the Current Task”
DESCRIPTION:Abstract:\nHumans often must respond quickly to events happening in their environment. To support fast perception and action\, the brain has evolved a warning system. Warning stimuli are used to elicit a transient state of readiness for perception and action (phasic alertness) that results in faster perceptual processing and faster decision-making for action. Phasic alertenss is assumed to be “unintelligent” in the sense that it is driven by the warning stimuli\, irrespective of the cognitive task set and the expectations guiding goal-directed behavior in the current task. Here\, we review recent findings that falsify this assumption. We provide evidence that phasic alertness presupposes an expectation that stimuli can serve as a warning within the current task. In addition\, we show that within a task\, phasic alertness unfolds in action-focused episodes that restrict its effects to only the next action in an action sequence. Together\, these findings reveal that phasic alertness is not entirely stimulus-based (bottom-up)\, but also relies on the cognitive mechanisms for (top-down) control of task-driven and goal-directed action and thus the “intelligent” interaction with the environment.\n\n\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Kirill Pershin on Unsplash \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-christian-poth-bielefeld-university-task-driven-phasic-alertness-how-being-ready-for-action-relies-on-the-current-task/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/kirill-pershin-U0BblJ-kQfA-unspl-scaled.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240222T100000
DTEND;TZID=Europe/Berlin:20240222T110000
DTSTAMP:20260615T153633
CREATED:20240205T134855Z
LAST-MODIFIED:20250603T125157Z
UID:18017-1708596000-1708599600@www.scienceofintelligence.de
SUMMARY:Tim Kietzmann (University of Osnabrück)\, “Large Language Models Offer a Rich Representational Format for Understanding the Transformation of Visual Information in the Human Brain.”
DESCRIPTION:Abstract: Originating from the connectionist movement of cognitive science\, deep neural networks (DNNs) have had tremendous influence on artificial intelligence\, operating at the core of today’s most powerful applications. At the same time\, cognitive computational neuroscientists have recognised their promise to act as “Goldilocks” models of brain function: DNNs are grounded in sensory data\, can be trained to perform complex tasks in a distributed fashion\, are fully configurable/accessible to the experimenter\, and can be mapped to brain function across various levels of explanation. This has led to a fruitful research cycle in which biological aspects are integrated into network design\, and the corresponding networks are then tested for their ability to predict neural and behavioural data. This talk will present this emerging approach\, which we call neuroconnectionism\, as a cohesive large-scale research programme centered around ANNs as a computational language for expressing falsifiable theories about brain computation. As a case study\, I will focus on a collaborative effort in which we test the ability of large-language models (LLMs) to provide a good representational format for modelling human visual responses to natural scenes. By running tightly controlled model comparisons\, we demonstrate that recurrent neural networks\, trained to map from pixels to semantic LLM embedding\, provide the current best account of a large-scale\, 7T fMRI dataset (NSD)\, outperforming other supervised as well as unsupervised ANN models. These findings point towards the view that vision may not be optimised for visual categorisation alone\, but instead maps from retinal input into a high-dimensional semantic format that can be captured by contextual learning in language.\n\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Pietro Jeng on Unsplash. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-tim-kietzmann-university-of-osnabruck-large-language-models-offer-a-rich-representational-format-for-understanding-the-transformation-of-visual-information-in-the-human-bra/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/pietro-jeng-n6B49lTx7NM-unsplash-1024x1024-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240201T100000
DTEND;TZID=Europe/Berlin:20240201T110000
DTSTAMP:20260615T153633
CREATED:20231113T091900Z
LAST-MODIFIED:20250603T125206Z
UID:17049-1706781600-1706785200@www.scienceofintelligence.de
SUMMARY:Stefan Leutgeb\, “Hippocampal Computations in Support of Spatial Navigation and Working Memory”
DESCRIPTION:Stefan Leutgeb is Professor of Neurobiology at University of California San Diego. Currently a fellow of the Wissenschaftskolleg zu Berlin with his research on neural computations in real brains and in artificial systems. More details to follow.\n\n\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Alina Grubnyak on Unsplash. \n  \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-stefan-leutgeb-hippocampal-computations-in-support-of-spatial-navigation-and-working-memory/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/11/alina-grubnyak-tEVGmMaPFXk-unsplash.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240125T100000
DTEND;TZID=Europe/Berlin:20240125T110000
DTSTAMP:20260615T153633
CREATED:20240108T132410Z
LAST-MODIFIED:20240415T104537Z
UID:18010-1706176800-1706180400@www.scienceofintelligence.de
SUMMARY:Oren Forkosh (The Hebrew University of Jerusalem)\, "Behavior\, Personality\, and Affective States of Freely Behaving Groups of Mice and Other Animals"
DESCRIPTION:Behavior\, Personality\, and Affective States of Freely Behaving Groups of Mice and Other Animals\nIn recent years\, the study of animal behavior in neuroscience has seen a significant shift towards more naturalistic and less intrusive methods. It is under these conditions that the true spectrum of animal behavior can be exhibited\, free from the artificial constraints and stressful conditions often imposed by traditional laboratory settings. In this talk\, I will discuss the interplay between behavior\, personality\, and affective states as measured in our “social boxes”; These systems allow for the continuous and unattended tracking of groups of mice over extended periods and can automatically recognize and catalog over 100 distinct behaviors. A four-day experiment\, for example\, can potentially replace a myriad of classical tests typically used in neuroscience. Our system can also discern and record a ‘behavioral fingerprint’ for each mouse. These fingerprints reveal consistent traits—personalities—that are not only distinct between individuals but also persist over time. In addition\, by examining the interplay between behavior and personality across multiple timescales – from seconds to days – we can gain insights into the affective states of these animals. Finally\, expanding our research to other species\, including bats\, cows\, and even humans\, allows us to develop a general understanding of behavior and personality. This comparative strategy holds promise for developing a ‘universal translator’ of behavioral and personality patterns\, paving the way for new comparative studies. These insights into the personalities and emotions of both humans and animals have the potential to significantly enhance our knowledge of the neurobiological underpinnings of behavior. \n—Oren Forkosh is a PI at the Lab for Computational Neuroscience- Behavior\, Personality and Cognition at the Hebrew University of Jerusalem. The Forkosh lab uses machine-learning to understand personality\, behaviour\, hierarchy\, communication\, social learning\, and much more\, in order to make the world happier for animals and people alike. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-oren-forkosh-the-hebrew-university-of-jerusalem-behavior-personality-and-affective-states-of-freely-behaving-groups-of-mice-and-other-animals/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/png:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/Screenshot-2024-01-08-at-10.35.11-300x166-1.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20240118T100000
DTEND;TZID=Europe/Berlin:20240118T110000
DTSTAMP:20260615T153633
CREATED:20231220T123724Z
LAST-MODIFIED:20250603T125238Z
UID:18004-1705572000-1705575600@www.scienceofintelligence.de
SUMMARY:Asieh Daneshi (Science of Intelligence)\, “Behavioral Contagion in Human and Artificial Multi-Agent Systems”
DESCRIPTION:In this talk\, Asieh will explore the dynamics of decision-making and risk-taking within social contexts and how everyday decisions\, often laden with potential negative outcomes\, are influenced not only by individual judgment but significantly by the surrounding social environment. Her research employs the “Balloon Analogue Risk-Taking” experiment in a controlled setting applying VR-technology in order to investigate the impact of dynamic social interactions on individual risk assessment. Various aspects\, such as the effect of peers on risk perception\, group dynamics in decision-making\, and the influence of social norms on risk-taking behaviors are part of her study. \nBy observing how individuals in a group adapt their decisions based on others’ actions and outcomes\, she aims to understand the balancing act between individual decision-making and group conformity. This research explores decision-making in a constantly changing social environment\, offering new insights into how group dynamics can lead to either more risk-taking or conservative behaviors. The findings promise to enhance our understanding of the complex interplay between personal psychology and group influence. \n\n\n\n\n\nThis talk will take place in person at SCIoI. \n  \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-asieh-daneshi-behavioral-contagion-in-human-and-artificial-multi-agent-systems/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2024/04/IMG_6596-1536x1536-1-1024x1024-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20231214T100000
DTEND;TZID=Europe/Berlin:20231214T110000
DTSTAMP:20260615T153633
CREATED:20231113T092531Z
LAST-MODIFIED:20240813T102801Z
UID:17050-1702548000-1702551600@www.scienceofintelligence.de
SUMMARY:Aravind Battaje and Vito Mengers\, "Principles at Play: What is Intelligence?"
DESCRIPTION:What is intelligence? We delve into the collaborative efforts at SCIoI\, where we aim to understand intelligence through the identification of commonalities. Inspired by ongoing research and historical parallels\, we present candidate principles\, inviting the audience to contribute insights and discuss their alignment with ongoing projects. This talk marks a step towards refining our understanding of intelligence\, emphasizing the pivotal role principles play in shaping our collective reflections and influencing future research trajectories.\n\n\n\n\nThis talk will take place in person at SCIoI. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-vito-mengers-project-35-differentiable-interconnected-recursive-estimation-as-a-principle-of-intelligence/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/11/Vito-Mengers-1-1536x1536-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20231130T100000
DTEND;TZID=Europe/Berlin:20231130T110000
DTSTAMP:20260615T153633
CREATED:20231020T092645Z
LAST-MODIFIED:20250603T125309Z
UID:17048-1701338400-1701342000@www.scienceofintelligence.de
SUMMARY:Eva Wiese (TU Berlin)\, “Social Perception and Attention in Human-Robot Interaction: Bottom-Up and Top-Down Influences”
DESCRIPTION:Eva Wiese is the professor for Cognitive Psychology and Ergonomics at TU Berlin. More details to follow.\n\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Possessed Photography on Unsplash\,  \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-eva-wiese/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/10/possessed-photography-zbLW0FG8XU8-unsplash-e1700583952116.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20231109T100000
DTEND;TZID=Europe/Berlin:20231109T110000
DTSTAMP:20260615T153633
CREATED:20231004T125227Z
LAST-MODIFIED:20250603T125349Z
UID:16860-1699524000-1699527600@www.scienceofintelligence.de
SUMMARY:Daniela Vallentin (MPI for Biological Intelligence)\, “Neural Mechanisms of Vocal Learning and Production in Songbirds”
DESCRIPTION:Daniela Vallentin is a neuroscientist and currently the Lise Meitner Reseach Group Leader at the Max Planck Institute for Biological Intelligence\, heading the ‘Neural Circuits for Vocal Communication’ Group whose objective is to explore the neural circuits driving skilled motor learning and orchestrating the coordination of precise movements by working with songbirds. Due to the homology of brain structures in birds and mammals\, studying the neural mechanisms of vocal learning and coordination in songbirds has the potential to reveal general principles of motor circuits in other animals\, including humans.\n\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Joshua J. Cotten on Unsplash \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-daniela-vallentin-mpi-for-biological-intelligence-neural-mechanisms-of-vocal-learning-and-production-in-songbirds/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/10/joshua-j-cotten-NgSPFaBWICo-unsp.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20231102T100000
DTEND;TZID=Europe/Berlin:20231102T110000
DTSTAMP:20260615T153633
CREATED:20230918T084845Z
LAST-MODIFIED:20250603T125402Z
UID:16719-1698919200-1698922800@www.scienceofintelligence.de
SUMMARY:Jonas Frenkel and Uroš Petković (Science of Intelligence)\, “Social Responsiveness and Its Effects on Learning in Human-Human and Human-Robot Interaction”
DESCRIPTION:More details to follow.\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Katja Anokhina on Unsplash. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-project-31-social-responsiveness-and-its-effects-on-learning-in-human-human-and-human-robot-interaction/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/09/katja-anokhina-_7ceGXTAtyQ-unsplash.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20231026T100000
DTEND;TZID=Europe/Berlin:20231026T110000
DTSTAMP:20260615T153633
CREATED:20230927T091538Z
LAST-MODIFIED:20250603T125413Z
UID:16730-1698314400-1698318000@www.scienceofintelligence.de
SUMMARY:Svetlana Levit\, “Analyzing Human Physical Reasoning and Strategy Exploration on Physical Puzzles”
DESCRIPTION:More details to follow.\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Markus Spiske on Unsplash. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-svetlana-levit-project-30/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/png:https://www.scienceofintelligence.de/wp-content/uploads/2023/09/Screen-Shot-2021-07-22-at-07.50.11.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20231019T100000
DTEND;TZID=Europe/Berlin:20231019T110000
DTSTAMP:20260615T153633
CREATED:20230918T084300Z
LAST-MODIFIED:20240813T102952Z
UID:16715-1697709600-1697713200@www.scienceofintelligence.de
SUMMARY:Richard Schweitzer (Science of Intelligence)\, "Preregistration in Open Science: What\, why\, and how (a live tutorial)"
DESCRIPTION:Abstract: \nA tutorial on Open Science practices with a focus on pre-registration\, going through the process step-by-step\, including a live experimental data collection.\n\n\nThis talk will take place in person at SCIoI. \nPhoto by Markus Spiske on Unsplash. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-richard-schweitzer-preregistration-in-open-science-what-why-and-how-a-live-tutorial/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/09/markus-spiske-70Rir5vB96U-unspla.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20231005T100000
DTEND;TZID=Europe/Berlin:20231005T110000
DTSTAMP:20260615T153633
CREATED:20230515T104937Z
LAST-MODIFIED:20250603T125727Z
UID:15430-1696500000-1696503600@www.scienceofintelligence.de
SUMMARY:Conor Heins\, “Collective Behavior From Surprise Minimization”
DESCRIPTION:Abstract: \nCollective motion is a familiar sight in nature; groups of distinct\, self-propelled individuals appear to move as a coherent whole\, exhibiting a rich behavioral repertoire that ranges from directed movement to milling to disordered swarming. Biological collective motion is an emergent phenomenon that is the result of self-organization\, whereby macroscopic patterns arise from decentralized\, local interactions among constituent components (e.g.\, individual fish in a school). Preminent models of collective motion describe individuals in the group as self-propelled particles\, subject to a combination of self-generated motion and “social forces” that depend on the state of neighboring particles. Here we introduce a fundamentally new approach to modelling collective movement in animal groups based on active inference\, a cognitive framework that casts behavior as consequences of a single imperative: to minimize surprise. We demonstrate that many empirically-observed collective phenomena\, including cohesion\, milling and directed motion\, naturally emerge when considering individual behavior as the consequence of active Bayesian inference — this emerges without ever explicitly building behavioral rules or goals into individual agents. We show that active inference can naturally recover and generalize the classical notion of social forces in agent-based models of collective motion. By analyzing the parameter space of the belief-based model\, we reveal non-trivial relationships between the beliefs of individuals and group properties like collective polarization and the probability of occupying different behavioral regimes. We also explore how individual beliefs about uncertainty influence the accuracy of collective decision-making. Finally\, we show how\, in this framework\, agents can readily update their generative model over time\, resulting in groups that are collectively more sensitive to external fluctuations and encode information more robustly.\n\n\nThis talk will take place in person at SCIoI. \nPhoto kinldy provided by Shintaro Shiba. \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-conor-heins-collective-behavior-from-surprise-minimization/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/05/sawyer-bengtson-umRPY9w3q1c-unsp.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20230720T100000
DTEND;TZID=Europe/Berlin:20230720T110000
DTSTAMP:20260615T153633
CREATED:20230605T103948Z
LAST-MODIFIED:20240813T103027Z
UID:15696-1689847200-1689850800@www.scienceofintelligence.de
SUMMARY:Lisa-Kristin Richter\, "Model Training for Facial Recognition of Raccoons"
DESCRIPTION:Machine learning tools have already been used to identify individual animals such as but not limited to pandas\, black bears\, cows and dogs. These tools can help to improve the quality of non-invasive wildlife monitoring and enhance the information on individual animal behaviour as well as on behaviour within social networks of the animals (Lynn 2019; Schofield et al. 2019). \nRaccoons (proctorloco) are considered an invasive species in Germany that has been introduced to many parts of the world outside of their native range in North America. \nIn order to train a model for facial recognition of raccoons\, we collected 7812 pictures of 133 individuals. After manual selection for quality focusing on sharpness\, image detail and light\, 111individuals with 4000 pictures remain in the dataset. The individuals were pictured in more than 10 different facilities and locations with different lights and angles. From this baseline dataset\, one data set using bounding boxes is created for training and one dataset using masks is also created for training. This is done to keep the influence of the background minimal. \nFinally\, this data is used to train different pre-trained deep learning models from Image Net\, namely ResNet50\, VGG19 and Mobile Net. While model training parameters like batch size\, number of epochs\, learning rate scheduler\, picture augmentation techniques and more are being varied. \nChallenges arise from the time and computer resources needed for training.Currently\, training is done via Google Colab\, which disconnects after a certain time. Furthermore\, input on dataset preprocessing\, model selection\, possible combination of models and variation in parameters would be very helpful. \nThis talk will take place in person at SCIoI. \nPhoto by Lukas Stoermer on Unsplash \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-lisa-kristin-richter/
LOCATION:MAR 2.057
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/06/lukas-stoermer-C0hX1cRQPOY-unspl.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20230713T100000
DTEND;TZID=Europe/Berlin:20230713T110000
DTSTAMP:20260615T153633
CREATED:20230605T103302Z
LAST-MODIFIED:20240813T103032Z
UID:15693-1689242400-1689246000@www.scienceofintelligence.de
SUMMARY:Santiago Paternain\, "Safe Learning for Dynamical Systems and Control"
DESCRIPTION:Abstract: \nReinforcement learning has shown great success in controlling complex dynamical systems. However\, when training a policy\, most algorithms only consider a single objective function. While this may suffice in virtual domains\, physical systems must satisfy a set of operational constraints\, with safety being of crucial importance. It is natural to express these problems as constrained optimization problems since weighted combinations of different rewards are not guaranteed to find a solution that satisfies all the requirements. Furthermore\, these examples are not contrived\, and safety-constrained reinforcement learning is a vital area of research that needs to be tackled. \nAfter establishing the need to tackle safety-constrained reinforcement learning\, I will shift my focus to solving these generally non-convex problems. I will discuss different approaches that exploit duality theory to pave the way towards algorithms for general constrained reinforcement learning. In particular\, I will discuss that (i) despite their non-convexity these problems have zero duality gap\, (ii) a state-augmented approach that does not guarantee convergence to an optimal policy but\, it guarantees optimality and (iii) a safe policy-gradient theorem that allows us to consider constraints beyond time-averages. \nThis talk will take place in person at SCIoI. \n  \nPhoto by Jeswin Thomas on Unsplash \n 
URL:https://www.scienceofintelligence.de/event/thursday-morning-talk-santiago-paternain/
CATEGORIES:Thursday Morning Talk
ATTACH;FMTTYPE=image/jpeg:https://www.scienceofintelligence.de/wp-content/uploads/2023/06/jeswin-thomas-dfRrpfYD8Iw-unspla.jpg
END:VEVENT
END:VCALENDAR