https://scholarreels.com/reelshttps://scholarreels.com/reels/abouthttps://scholarreels.com/reels/teamhttps://scholarreels.com/reels/privacyhttps://scholarreels.com/reels/termshttps://scholarreels.com/reels/policyhttps://scholarreels.com/reels?author=Alperen+Degirmenci2026-10-24https://scholarreels.com/reels?author=Christos+Kozyrakis2026-10-24https://scholarreels.com/reels?author=Hugo+Hadfield2026-10-24https://scholarreels.com/reels?author=Jason+Clemons2026-10-24https://scholarreels.com/reels?author=Rowland+O'Flaherty2026-10-24https://scholarreels.com/reels?author=Shuran+Song2026-10-24https://scholarreels.com/reels?author=Syed+Eqbal+Alam2026-10-24https://scholarreels.com/reels?author=Wenqi+Jiang2026-10-24https://scholarreels.com/reels?author=Yashraj+Narang2026-10-24https://scholarreels.com/reels?author=Zhan+Shu2026-10-24https://scholarreels.com/reels?cat=cs.AI2026-10-24https://scholarreels.com/reels?cat=cs.RO2026-10-24https://scholarreels.com/reels?institution=NVIDIA+Research2026-10-24https://scholarreels.com/reels?institution=SheQAI+Research2026-10-24https://scholarreels.com/reels?institution=Stanford+University2026-10-24https://scholarreels.com/reels?institution=University+of+Alberta2026-10-24https://scholarreels.com/reels?tag=AI+agents2026-10-24https://scholarreels.com/reels?tag=distributed+optimization2026-10-24https://scholarreels.com/reels?tag=federated+multi-agent+system2026-10-24https://scholarreels.com/reels?tag=generative+AI2026-10-24https://scholarreels.com/reels?tag=multimodal+foundation+models2026-10-24https://scholarreels.com/reels?tag=optimal+control2026-10-24https://scholarreels.com/reels?tag=Robotics2026-10-24https://scholarreels.com/reels?tag=Robotics+Oriented+Serving+Arch2026-10-24https://scholarreels.com/reels?tag=ROSA2026-10-24https://scholarreels.com/reels?type=abstracts2026-10-24https://scholarreels.com/reels?type=highlights2026-10-24https://scholarreels.com/reels?type=shorts2026-10-24https://scholarreels.com/reels/u/drsyed2026-10-24https://scholarreels.com/reels/org/sheqai-researchhttps://scholarreels.com/reels/69ed6a4231b14d0abdff427ec9a5b7832026-10-24https://scholarreels.com/reels/69ed6a4231b14d0abdff427ec9a5b783/media/poster?v=76da9b9429A field experiment of social influence and behavioral contagion with bots on RedditRecent advances in AI have heightened scholars' and policy makers' concern with social influence and behavioral contagion in online communities. We conduct a field experiment on Reddit to investigate the extent to which online users are susceptible to positive behavioral stimuli from other users and artificial agents. We let apparent human and bot accounts give symbolic awards to users with one of four rationales: praising the recipient's logical argument, emotional sensitivity, or moral integrity, or explaining that the award resulted from a random draw in a lottery. We evaluate how the different rationales for the award affect the recipients' subsequent behavior on the platform in terms of volume, impact, and content, as well as the further behavioral contagion to other users.https://scholarreels.com/reels/69ed6a4231b14d0abdff427ec9a5b783/media/socialhttps://scholarreels.com/reels/69ed6a4231b14d0abdff427ec9a5b783/embed552026-07-05T09:34:47+00:00https://scholarreels.com/reels/a364ec770ff6433d91e5df10de96a2202026-10-24https://scholarreels.com/reels/a364ec770ff6433d91e5df10de96a220/media/poster?v=d93be2fe09Decoding Alignment without Encoding Alignment: A critique of similarity analysis in neuroscienceRSA and CKA are standard metrics for comparing neural representations across brain regions, organisms, and deep learning models. We demonstrate a fundamental weakness: these decoding-based metrics are insensitive to encoding manifold topology---the internal functional organization of a neural population. In a controlled MNIST experiment, RSA, CKA, and Procrustes R2 remain statistically unchanged when encoding topology is causally manipulated via an auxiliary clustering loss, while the two model populations differ significantly in attribution patterns, weight-graph assortativity, and out-of-distribution robustness. Across biological systems and machine learning models, similar decoding behavior can arise from small, non-representative subpopulations, and alignment metrics are insensitive to encoding manifold topology even when it is fundamentally altered.https://scholarreels.com/reels/a364ec770ff6433d91e5df10de96a220/media/socialhttps://scholarreels.com/reels/a364ec770ff6433d91e5df10de96a220/embed592026-07-06T05:13:15+00:00https://scholarreels.com/reels/8d915933ecf24967b062f405e424d9052026-10-24https://scholarreels.com/reels/8d915933ecf24967b062f405e424d905/media/poster?v=71aa4e1fb0Swin Transformer: Hierarchical Vision Transformer using Shifted WindowsThis paper presents a new vision Transformer, called Swin Transformer, that capably serves as a general-purpose backbone for computer vision. Challenges in adapting Transformer from language to vision arise from differences between the two domains, such as large variations in the scale of visual entities and the high resolution of pixels in images compared to words in text. To address these differences, we propose a hierarchical Transformer whose representation is computed with Shifted windows. The shifted windowing scheme brings greater efficiency by limiting self-attention computation to non-overlapping local windows while also allowing for cross-window connection. This hierarchical architecture has the flexibility to model at various scales and has linear computational complexity with respect to image size.https://scholarreels.com/reels/8d915933ecf24967b062f405e424d905/media/socialhttps://scholarreels.com/reels/8d915933ecf24967b062f405e424d905/embed602026-07-06T05:19:26+00:00https://scholarreels.com/reels/1fcb069e3f1e44998d09ac7c68a7676e2026-10-24https://scholarreels.com/reels/1fcb069e3f1e44998d09ac7c68a7676e/media/poster?v=d108165693ROSA: A Robotics Foundation Model Serving System for Robot FactoriesRobotics foundation models (RFMs) are making general-purpose robots increasingly practical for factory deployments. While RFM serving systems are central to this vision, existing systems are largely shaped by a single-robot, single-model assumption: inference is treated as an edge-computing problem handled by an on-robot or dedicated nearby GPU, and the serving objective is to minimize the latency of a single action model. In this paper, we propose ROSA, an RFM serving system for robot factories designed around three key principles. First, ROSA adopts shared GPU-pool serving, allowing a fleet of robots to access powerful server-class GPUs over the network in order to improve inference performance, battery duration, and GPU utilization.https://scholarreels.com/reels/1fcb069e3f1e44998d09ac7c68a7676e/media/mainhttps://scholarreels.com/reels/1fcb069e3f1e44998d09ac7c68a7676e/embed752026-07-07T07:37:28+00:00https://scholarreels.com/reels/f7084094bef74255b93c617a1593c58f2026-10-24https://scholarreels.com/reels/f7084094bef74255b93c617a1593c58f/media/poster?v=98409c6ee2ROSA: A Robotics Foundation Model Serving System for Robot FactoriesRobotics foundation models (RFMs) are making generalpurpose robots increasingly practical for factory deployments. While RFM serving systems are central to this vision, existing systems are largely shaped by a single-robot, single-model assumption: inference is treated as an edge-computing problem handled by an on-robot or dedicated nearby GPU, and the serving objective is to minimizing the latency of a single action model. In this paper, we propose ROSA, the Robotics Oriented Serving Architecture, an RFM serving system for robot factories designed around three key principles. First, ROSA adopts shared GPU-pool serving, allowing a fleet of robots to access powerful server-class GPUs over the network in order to improve inference performance, battery duration, and GPU utilization.https://scholarreels.com/reels/f7084094bef74255b93c617a1593c58f/media/socialhttps://scholarreels.com/reels/f7084094bef74255b93c617a1593c58f/embed412026-07-07T07:37:55+00:00https://scholarreels.com/reels/4fba57056b294ef9b28d2738acc2ab492026-10-24https://scholarreels.com/reels/4fba57056b294ef9b28d2738acc2ab49/media/poster?v=5bb4d643ffCollaborative AI Agents and Critics for Fault Detection and Cause Analysis in Network TelemetryWe develop algorithms for collaborative control of AI agents and critics in a multi-actor, multi-critic federated multi-agent system. Each AI agent and critic has access to classical machine learning or generative AI foundation models. The AI agents and critics collaborate with a central server to complete multimodal tasks such as fault detection, severity, and cause analysis in a network telemetry system, text-to-image generation, video generation, healthcare diagnostics from medical images and patient records, etcetera. The AI agents complete their tasks and send them to AI critics for evaluation. The critics then send feedback to agents to improve their responses. Collaboratively, they minimize the overall cost to the system with no inter-agent or inter-critic communication. AI agents and critics keep their cost functions or derivatives of cost functions private.https://scholarreels.com/reels/4fba57056b294ef9b28d2738acc2ab49/media/mainhttps://scholarreels.com/reels/4fba57056b294ef9b28d2738acc2ab49/embed1742026-07-06T06:54:03+00:00https://scholarreels.com/reels/b767bf48185649e390a0f1819260c56d2026-10-24https://scholarreels.com/reels/b767bf48185649e390a0f1819260c56d/media/poster?v=404148d5e0Superintelligent Agents Pose Catastrophic Risks: Can Scientist AI Offer a Safer Path?The leading AI companies are increasingly focused on building generalist AI agents---systems that can autonomously plan, act, and pursue goals across almost all tasks that humans can perform. Despite how useful these systems might be, unchecked AI agency poses significant risks to public safety and security, ranging from misuse by malicious actors to a potentially irreversible loss of human control. We discuss how these risks arise from current AI training methods. Indeed, various scenarios and experiments have demonstrated the possibility of AI agents engaging in deception or pursuing goals that were not specified by human operators and that conflict with human interests, such as self-preservation. Following the precautionary principle, we see a strong need for safer, yet still useful, alternatives to the current agency-driven trajectory.https://scholarreels.com/reels/b767bf48185649e390a0f1819260c56d/media/socialhttps://scholarreels.com/reels/b767bf48185649e390a0f1819260c56d/embed282026-07-21T23:42:36+00:00https://scholarreels.com/reels/sub-32026-10-24https://scholarreels.com/reels/sub-3/media/poster?v=31b7d06645Existence of a Unique Invariant Measure and Ergodic Property in AIMD-based Multi-resource AllocationDistributed resource allocation arises in many application domains, such as smart energy systems, intelligent transportation systems, cloud computing, edge computing, etcetera. To realize many of these applications, agents in a network may require multiple shared resources to complete a task and aim to maximize the network utility. Additionally, they may demand resources based on their preferences. Furthermore, they may not wish to share their cost functions, partial derivatives of the cost functions, etc., with other agents or a central server; however, they share their resource demands with the central server that aggregates the demands and sends one-bit resource-capacity constraint notification in the network.https://scholarreels.com/reels/sub-3/media/mainhttps://scholarreels.com/reels/sub-3/embed1682026-08-17T01:49:21+00:00https://scholarreels.com/reels/8e6cf356fd5842c6a0a3a2c6a5e1929e2026-10-24https://scholarreels.com/reels/8e6cf356fd5842c6a0a3a2c6a5e1929e/media/poster?v=97b36d8c0cPrint-Aware Synthesis and Physical Design Methodologies for 3D-Printed Microfluidic BiochipsMicrofluidic devices are widely used in diagnostics, chemical synthesis, and biological analysis, but their development often depends on complex fabrication and design processes. Resin-based three-dimensional (3D) printing has emerged as a promising alternative to conventional microfabrication because it enables low-cost, rapid prototyping of complex multi-layer structures. However, the practical realization of 3D-printed microfluidic biochips remains challenging due to manual and expertise-intensive design workflows, the rigid nature of commonly used printing materials, and fabrication inaccuracies such as over-curing that distort internal features and may block narrow channels. In this paper, we present a cohesive design automation framework for 3D-printed microfluidics that addresses these challenges across both device design and fabrication.https://scholarreels.com/reels/8e6cf356fd5842c6a0a3a2c6a5e1929e/media/socialhttps://scholarreels.com/reels/8e6cf356fd5842c6a0a3a2c6a5e1929e/embed452026-09-08T17:32:55+00:00https://scholarreels.com/reels/sub-52026-10-24https://scholarreels.com/reels/sub-5/media/poster?v=802d857422Collaborative AI Agents and Critics for Fault Detection and Cause Analysis in Network TelemetryWe develop algorithms for collaborative control of AI agents and critics in a multi-actor, multi-critic federated multi-agent system. Each AI agent and critic has access to classical machine learning or generative AI foundation models. The AI agents and critics collaborate with a central server to complete multimodal tasks such as fault detection, severity, and cause analysis in a network telemetry system, text-to-image generation, video generation, healthcare diagnostics from medical images and patient records, etcetera. The AI agents complete their tasks and send them to AI critics for evaluation. The critics then send feedback to agents to improve their responses. Collaboratively, they minimize the overall cost to the system with no inter-agent or inter-critic communication. AI agents and critics keep their cost functions or derivatives of cost functions private.https://scholarreels.com/reels/sub-5/media/mainhttps://scholarreels.com/reels/sub-5/embed742026-09-08T20:58:04+00:00https://scholarreels.com/reels/1357264586bc4a068951d73606eb13112026-10-24https://scholarreels.com/reels/1357264586bc4a068951d73606eb1311/media/poster?v=82e5a843ceOverLay++: Dense-Overlap Layout-to-Image Generation DatasetLayout-to-Image generation has made substantial progress in spatial and object-level control. However, existing methods still struggle with complex scenes containing many overlapping and interacting objects. We argue that training data is a particular bottleneck: existing datasets lack examples with dense, complex object interactions. To address this gap, we introduce , a large-scale Layout-to-Image dataset with structurally complex scenes. contains approximately 500K images with an average of 6.6 objects per image, exceeding existing datasets by in annotation density. Beyond annotation density, provides rich semantic detail with object captions over six times longer than in current datasets. Our dataset generation pipeline is simple and produces dense, overlapping object annotations with rich per-object captions.https://scholarreels.com/reels/1357264586bc4a068951d73606eb1311/media/socialhttps://scholarreels.com/reels/1357264586bc4a068951d73606eb1311/embed412026-10-13T00:47:09+00:00