Other🚀 Author shipped product
80
Experimental Performance of a 5G N78 Reconfigurable Intelligent Surface: From Controlled Measurements to Commercial Network Deployment
Sefa Kayraklık, Samed Keşir, Batuhan Kaplan, Ahmet Muaz Aktaş
·
Academic Institution
·
30 Apr 2026
This paper addresses the problem of enhancing coverage in non-line-of-sight zones using a reconfigurable intelligent surface and presents a real-world experimental analysis of a modular prototype. The results show promising gains in RSRP and SINR, indicating improved coverage and signal quality.
BUSINESS OPPORTUNITY
A concrete business opportunity for this technology is to offer network coverage enhancement services to telecommunications companies or businesses in areas with poor network coverage, using the reconfigurable intelligent surface to improve signal quality and strength. This could be sold as a subscription-based service, with the business opportunity grounded in the experimental results demonstrating improved coverage and signal quality.
Computing Equilibrium beyond Unilateral Deviation
Mingyang Liu, Gabriele Farina, Asuman Ozdaglar
·
Academic Institution
·
30 Apr 2026
This paper addresses the problem of equilibrium concepts in game theory, specifically the limitation of familiar equilibrium concepts such as Nash and correlated equilibrium, and studies an alternative solution concept that minimizes coalitional deviation incentives. The authors propose a solution concept that minimizes coalitional deviation incentives, rather than requiring them to vanish, but do not explicitly state the outcome or result of this concept.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this research could be to develop and sell a software tool that helps organizations design and analyze coalition-proof equilibrium concepts, which could be useful in a variety of fields, including economics, politics, and cybersecurity. This tool could be marketed to consulting firms, think tanks, and other organizations that need to analyze and mitigate the risks of coalitional deviations.
PhyCo: Learning Controllable Physical Priors for Generative Motion
Sriram Narayanan, Ziyu Jiang, Srinivasa Narasimhan, Manmohan Chandraker
·
Academic Institution
·
30 Apr 2026
This paper addresses the problem of physical consistency in video diffusion models and presents PhyCo, a framework that introduces continuous, interpretable, and physically grounded control into video generation. The authors propose a framework that integrates three key components, including a large-scale dataset of photorealistic simulation videos and physics-supervised fine-tuning, but do not explicitly state the results or outcomes of their approach.
BUSINESS OPPORTUNITY
A concrete business opportunity for this research is to license the PhyCo framework to companies that specialize in video game development or special effects, where realistic physics-based video generation is a key requirement. This could be offered as a software development kit or API integration, allowing these companies to enhance their video generation capabilities.
DEFault++: Automated Fault Detection, Categorization, and Diagnosis for Transformer Architectures
Sigma Jahan, Saurabh Singh Rajput, Tushar Sharma, Mohammad Masudur Rahman
·
ETH
·
30 Apr 2026
This paper addresses the problem of faults in transformer models and presents DEFault++, a hierarchical learning-based diagnostic technique. The authors propose DEFault++ as a solution to detect, classify, and diagnose faults in transformer architectures, but do not explicitly state the outcome or results.
BUSINESS OPPORTUNITY
A concrete business opportunity for this research is to offer DEFault++ as a diagnostic tool for AI model faults to businesses that rely on transformer architectures. This could be sold as a standalone product or integrated into existing AI development platforms, but the abstract does not provide explicit details on commercialization.
Splitting Argumentation Frameworks with Collective Attacks and Supports
Matti Berthold, Lydia Blümel, Giovanni Buraglio, Anna Rapberger
·
Academic Institution
·
30 Apr 2026
This paper addresses the problem of argumentation formalisms and proposes novel splitting techniques for bipolar set-based argumentation frameworks, which incorporate supports between defeasible elements. The result or outcome is the establishment of a crucial link to structured argumentation, naturally capturing general assumption-based argumentation.
BUSINESS OPPORTUNITY
A concrete business opportunity grounded in this paper is the development of an argumentation-based decision-support tool, which could be licensed to organizations that regularly engage in complex decision-making processes. This tool could be tailored to specific industries, such as legal or finance, where the ability to effectively argue and resolve disputes is crucial.
Mapping the Methodological Space of Classroom Interaction Research: Scale, Duration, and Modality in an Age of AI
Dorottya Demszky, Edith Bouton, Alison Twiner, Sara Hennessy
·
ETH
·
30 Apr 2026
This paper addresses the methodological space of classroom interaction research and proposes a framework mapping this space along three dimensions--scale, duration, and modality. The authors illustrate this framework through contrasting studies of dialogic teaching and an interview with the lead researchers, but do not explicitly state a specific result or outcome.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper could be to develop an educational software that analyzes teacher-student interactions, however, the abstract does not provide enough information to ground this opportunity. The paper's focus on classroom interaction research does not explicitly demonstrate a clear business opportunity, but it could potentially be adapted for commercial use in the educational sector.
3D Reconstruction Techniques in the Manufacturing Domain: Applications, Research Opportunities and Use Cases
Chialoon Cheng, Kaijun liu, Zhiyang Liu, Marcelo H Ang
·
ETH
·
30 Apr 2026
This paper addresses the evolution and current state of three-dimensional (3D) reconstruction techniques in manufacturing applications, using traditional approaches and emerging deep learning methods. The authors classify reconstruction techniques into three primary categories: data acquisition, point cloud generation, post-processing and applications, and identify a critical research gap in unified 3D reconstruction frameworks.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper could be to offer a 3D reconstruction service to manufacturing businesses, providing them with detailed 3D models of their facilities or equipment. This service could be sold as a one-time project or as an ongoing subscription-based model, with the goal of improving the customer's asset management and monitoring capabilities.
RHyVE: Competence-Aware Verification and Phase-Aware Deployment for LLM-Generated Reward Hypotheses
Feiyu Wu, Xu Zheng, Zhuocheng Wang, Yi ming Dai
·
Academic Institution
·
30 Apr 2026
This paper addresses the problem of verifying and deploying reward hypotheses generated by large language models in reinforcement learning, using a method called RHyVE. The authors propose RHyVE, but the abstract does not explicitly state the outcome or result of this proposal.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper could be to offer consulting services to businesses on how to improve the reliability of their reinforcement learning-based automation systems using RHyVE. This could involve providing tailored solutions and workflows to clients, but the abstract does not provide enough information to determine the specifics of such an opportunity.
To Build or Not to Build? Factors that Lead to Non-Development or Abandonment of AI Systems
Shreya Chappidi, Jatinder Singh
·
MIT
·
30 Apr 2026
This paper addresses the problem of non-development or abandonment of AI systems and uses a scoping review of academic literature to investigate factors influencing these decisions. The authors do not explicitly state the outcome or result of their investigation in the abstract.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper could be to offer consulting services to businesses on how to avoid non-development or abandonment of AI systems. This could be sold as a premium service to businesses looking to improve their AI development efficiency.
SpecVQA: A Benchmark for Spectral Understanding and Visual Question Answering in Scientific Images
Jialu Shen, Han Lyu, Suyang Zhong, Hanzheng Li
·
Academic Institution
·
30 Apr 2026
This paper addresses the challenge of evaluating multimodal large language models on scientific spectral understanding and introduces SpecVQA, a benchmark for spectral understanding and visual question answering in scientific images. The result is a benchmark containing 620 figures and 3100 question-answer pairs, covering 7 representative spectrum types with expert-annotated question-answer pairs.
BUSINESS OPPORTUNITY
A concrete business opportunity for this research is to license the SpecVQA benchmark to companies or institutions that work with scientific spectral data, providing them with a standardized evaluation tool for their multimodal models. This could be sold as a subscription-based service, with the benchmark being updated regularly to include new spectrum types and question-answer pairs.
Design Structure Matrix Modularization with Large Language Models
Shuo Jiang, Jianxi Luo
·
ETH
·
30 Apr 2026
This paper addresses the problem of Design Structure Matrix (DSM) modularization using Large Language Models (LLMs), building on prior work on LLM-based combinatorial optimization for DSM sequencing. The authors' method achieves near-reference quality within 30 iterations without requiring specialized optimization code.
BUSINESS OPPORTUNITY
A concrete business opportunity for this research is to offer LLM-based system modularization as a service to companies that struggle with complex system design, providing a more efficient and effective solution. This service could be sold as a consultancy package, where the LLM-based tool is used to analyze and optimize the client's system design.
A Pattern Language for Resilient Visual Agents
Habtom Kahsay Gidey, Alexander Lenz, Alois Knoll
·
Academic Institution
·
30 Apr 2026
This paper addresses the challenge of integrating multimodal foundation models into enterprise ecosystems by proposing an architectural pattern language for visual agents that separates fast, deterministic reflexes from slow, probabilistic supervision. The authors propose four architectural design patterns, including Hybrid Affordance Integration, but do not explicitly state the outcome or results of this proposal.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper could be to offer consulting services to help businesses integrate multimodal foundation models into their enterprise ecosystems. This could involve designing and implementing custom architectural pattern languages for visual agents to improve the performance and determinism of their systems.
Exploring Interaction Paradigms for LLM Agents in Scientific Visualization
Jackson Vonderhorst, Kuangshi Ai, Haichao Miao, Shusen Liu
·
Academic Institution
·
30 Apr 2026
This paper addresses the problem of using large language model (LLM) agents for scientific visualization tasks by comparing different interaction paradigms. The authors evaluate eight representative agents across 15 benchmark tasks and measure visualization quality, efficiency, robustness, and computational cost.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper could be to offer customized scientific visualization services to research institutions, using the evaluated LLM agents to generate high-quality visualizations. This service could be sold as a subscription-based model, providing access to a range of visualization tools and workflows.
ITS-Mina: A Harris Hawks Optimization-Based All-MLP Framework with Iterative Refinement and External Attention for Multivariate Time Series Forecasting
Pourya Zamanvaziri, Amirhossein Sadr, Aida Pakniyat, Dara Rahmati
·
Academic Institution
·
30 Apr 2026
This paper addresses the problem of multivariate time series forecasting and proposes a novel all-MLP framework called ITS-Mina that integrates an iterative refinement mechanism and external attention. The authors propose this framework as a potentially competitive or superior alternative to Transformer-based architectures with reduced computational cost, but do not explicitly state the outcome or result of using ITS-Mina.
BUSINESS OPPORTUNITY
A concrete business opportunity grounded in this paper is to offer ITS-Mina-based forecasting services to businesses that rely heavily on accurate time series forecasting, such as energy or financial institutions. This service could be sold as a subscription-based model, where clients receive regular forecasting updates and insights to inform their business decisions.
From LLM-Driven Trading Card Generation to Procedural Relatedness: A Pokémon Case Study
Johannes Pfau, Panagiotis Vrettis
·
Academic Institution
·
30 Apr 2026
This paper addresses the problem of predictable strategies and diminished viable card options in Trading Card Games by using Large Language Models and Image Diffusion Models for Procedural Content Generation of TCG cards. The authors investigate the use of these AI methods to enable a persona, but do not explicitly state the outcome or result of this investigation.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper is to offer AI-generated trading card content to online gaming platforms, increasing player engagement and reducing design costs. This could be sold as a subscription-based service, providing a constant stream of new and unique digital trading cards.
Graph World Models: Concepts, Taxonomy, and Future Directions
Jiawei Liu, Senqiao Yang, Mingjun Wang, Yu Wang
·
MIT
·
30 Apr 2026
This paper addresses the limitations of classical world models, including noise sensitivity and weak reasoning, by using graph structure to decompose the environment into entity nodes and interactive edges. The authors systematically formalize and unify emerging graph-based works, but do not explicitly state any results or outcomes.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper could involve developing and licensing graph-based world models to companies that operate complex systems, such as manufacturing or logistics firms. This could involve partnering with system integrators or IT consultants to deploy and customize the models for specific client use cases.
Robustness Evaluation of a Foundation Segmentation Model Under Simulated Domain Shifts in Abdominal CT: Implications for Health Digital Twin Deployment
Sanghati Basu
·
Academic Institution
·
28 Apr 2026
This paper addresses the robustness of the Segment Anything Model (SAM) for spleen segmentation in abdominal CT under simulated domain shifts, using a systematic slice-level robustness audit. The authors present a standardized ground-truth-derived bounding-box protocol to isolate encoder robustness from prompt uncertainty, but do not explicitly state the outcome or results.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper could involve developing and licensing the standardized ground-truth-derived bounding-box protocol for use in medical image analysis. This protocol could be sold to medical imaging software companies, which could then integrate it into their products to improve accuracy and robustness.
Feasible-First Exploration for Constrained ML Deployment Optimization in Crash-Prone Hierarchical Search Spaces
Christian Lysenstøen
·
MIT
·
27 Apr 2026
This paper addresses the problem of deploying machine learning models under production constraints and uses a method called Feasible-First Exploration for Constrained ML Deployment Optimization. The authors discuss the limitations of standard black-box optimizers in handling invalid configurations, but do not explicitly state the outcome or result of their proposed method.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper could be to offer a service that optimizes machine learning model deployment for clients with constrained environments. This service could be targeted at businesses that rely heavily on machine learning, such as healthcare or finance, and could be sold as a consulting package or a software solution.
Network Impact of Post-Quantum Certificate Chain sizes on Time to First Byte in TLS Deployments
Matthew Chou, Phuong Cao
·
TUM
·
27 Apr 2026
This paper addresses the problem of transitioning existing internet infrastructure to quantum-safe certificate chains and the impact on time to first byte in TLS deployments, using an evaluation method under CDN-focused TLS conditions. The authors explicitly state the goal of characterizing the latency cost of this transition, but do not mention specific outcomes or results.
BUSINESS OPPORTUNITY
A concrete business opportunity based on this paper is to offer consulting services to help businesses optimize their TLS deployments for quantum-safe certificate chains. This service could be targeted at companies that rely heavily on online transactions and need to ensure secure and efficient data transfer.
Characterizing Vision-Language-Action Models across XPUs: Constraints and Acceleration for On-Robot Deployment
Kaijun Zhou, Qiwei Chen, Da Peng, Zhiyang Li
·
Academic Institution
·
27 Apr 2026
This paper addresses the problem of deploying Vision-Language-Action (VLA) models on robots under tight cost and energy budgets using model-hardware co-characterization. The authors present a systematic analysis and build a cross-accelerator leaderboard to evaluate model-hardware pairs under Cost, Energy, Time (CET) metrics.
BUSINESS OPPORTUNITY
A concrete business opportunity for this research is to provide VLA model deployment services to robotics manufacturers, helping them optimize their models for edge devices and reduce costs. This service could be sold as a consulting package, where the researchers work with the manufacturer to characterize and optimize their VLA models for deployment on robotic devices.