Research & Education

Zero And Few Shot Reasoning is an AI tool hosted on Hugging Face, designed to explore the capabilities of large language models in solving complex problems. Users can input riddles or mathematical problems and receive solutions generated by various language models. A built-in calculator is available to help verify the accuracy of the provided answers. This tool is particularly useful for researchers and educators interested in understanding and evaluating the performance of AI in zero-shot and few-shot reasoning tasks, offering a practical environment for experimentation and analysis.

Yoloe is an AI application hosted on Hugging Face Spaces, developed by jameslahm (Ao Wang). This tool specializes in object detection and segmentation within images. Users can upload an image and then utilize various methods to identify and isolate objects. These methods include providing text descriptions, drawing bounding boxes, or creating masks. Additionally, Yoloe offers a prompt-free mode for more automated detection. It's designed to be accessible and provides a platform for exploring computer vision capabilities.

CrossQuery is a Chrome extension designed to significantly enhance the Google search experience. It integrates summaries from Reddit and other user-defined websites directly into a sidebar alongside standard Google search results. This eliminates the need to manually add 'site:reddit.com' to queries, streamlining the research process. The tool is ideal for users who frequently seek community-driven insights or specific information from particular sources, allowing for quicker comprehension and more focused information gathering. It aims to provide a more efficient way to access diverse perspectives and summarized content, making it easier to grasp key insights and understand complex topics.

HorizonX is an AI-native research workspace designed for ML engineers and researchers, offering a comprehensive platform to streamline the research process. It provides access to over 300 million research papers, ensuring zero hallucinations with verified citations. The tool features agentic brainstorming, allowing users to ask questions and receive structured, cited answers grounded in real research. Its literature review capability pinpoints relevant papers, while the 'Chat with PDF' function enables conversational interaction with single or multiple documents. HorizonX also includes an integrated writing canvas that provides full context from research, along with inline AI editing tools to sharpen, formalize, or restructure text without switching tabs. This end-to-end solution aims to accelerate research from question to published-grade work.

Learning Collider is a nonprofit research lab dedicated to building cutting-edge AI solutions that bridge academia and product development. The organization focuses on driving economic mobility and social impact through rigorous, evidence-based decision technology. They collaborate with mission-driven for-profit and nonprofit organizations to develop AI solutions that align with product roadmaps while prioritizing fairness and evidence. Learning Collider attracts top-tier data and social scientists to conduct research, deliver valuable R&D to partners, and generate measurable social impact for historically underrepresented groups. Their work emphasizes building AI with scientific rigor, optimized for social impact, rather than solely focusing on research output.

JuristAI offers an invisible legal AI solution designed for solo practitioners, small firms, and pro se litigants, integrating directly into email workflows. This superintelligent agent tracks dockets 24/7 and is practice-area and jurisdiction-specific. It automates critical legal tasks such as research, routine drafting, docket ingestion, client onboarding, deadline management, and client correspondence. By working seamlessly in the background, JuristAI aims to streamline operations and enhance efficiency for legal professionals without requiring a separate platform.

iHub-Drishti, promoted by IIT Jodhpur under a National Mission on Interdisciplinary Cyber Physical Systems (NM-ICPS), is a Technology Innovation Hub specializing in Computer Vision, Augmented Reality, and Virtual Reality. This not-for-profit organization aims to create a robust ecosystem by fostering collaborations with national and international institutions, domain experts, and industry partners including startups. The hub hosts faculty members, research scholars, developers, and scientists dedicated to advancing these technologies. It also focuses on building state-of-the-art infrastructure and a network of researchers to become a leading center for CV, AR, and VR technologies in India.

The Consejo Superior de Investigaciones Científicas (CSIC) is a Spanish State Agency dedicated to scientific research and technological development. IIIA-CSIC, as part of this larger institution, focuses specifically on AI research, balancing fundamental scientific inquiry with practical applications. The institute is committed to training PhD students and facilitating technology transfer, ensuring that research findings contribute to societal and industrial advancement. Their work spans various scientific areas, structured into global areas like Society, Life, and Matter, covering a broad spectrum of human knowledge. CSIC also emphasizes open science, international collaboration, and scientific dissemination to make science accessible to a wider audience.

Typeblock is a comprehensive tool designed for research and data analysis, catering to a wide range of methodologies including quantitative, qualitative, and mixed methods research. It assists users in efficiently analyzing complex datasets and identifying meaningful correlations between various variables. Beyond analysis, Typeblock aids in summarizing research findings clearly and concisely, making it easier to communicate results. The tool also provides support for suggesting future research directions, helping researchers to build upon their current work and explore new avenues. Its capabilities make it suitable for academic, professional, and personal research endeavors.

ViCOROB (Vision for Computer and Robotics) is a research group based at the University of Girona, established in 1993. The group focuses on multidisciplinary research in computer vision and robotics, integrating staff with diverse academic backgrounds. ViCOROB is dedicated to both fundamental and applied research, aiming to advance academic knowledge and facilitate technology transfer through collaborative projects. With a team of approximately 70 people, it contributes significantly to the fields of computer vision and robotics, fostering innovation and development within the university and beyond.

Vicena, formerly known as MirrorThink, is an advanced AI research assistant designed for scientists and engineers. It automates literature reviews across PubMed, arXiv, Google Scholar, and Google Patents, using adaptive query refinement for comprehensive coverage. The tool can extract detailed, variable-by-variable experimental protocols from scientific papers, assigning confidence scores based on source agreement. A key feature is its virtual lab simulation, which allows users to run and validate protocols against thermodynamic, kinetic, and chemical constraints before physical experimentation. Vicena also offers a science computer with an agentic JupyterLab environment, pre-installed with over 30 scientific packages, enabling collaborative work between users and the AI. Additionally, it provides chemistry tools for compound analysis, reaction prediction, and quantum chemistry calculations, all grounded in real physics and chemistry.

UMBRELLA_IHI is a groundbreaking initiative transforming stroke care throughout Europe by integrating AI and cutting-edge digital technologies. This project, a collaboration of over 20 public and private organizations led by Vall d’Hebron Institute of Research (VHIR) and Siemens Healthineers, aims to enhance stroke prevention, diagnosis, and reduce time to treatment, ultimately preventing long-term damage. It establishes a new paradigm in stroke management by standardizing real-world stroke data, developing a federated data platform (U-platform) for AI algorithm validation, and utilizing a federated learning system (FL-platform) to ensure data security and privacy while enabling collective AI training. The initiative also focuses on facilitating communication across interoperable silos for patient engagement, real-time monitoring, and decision-making, alongside developing a roadmap for regulatory compliance.

AI News Online offers daily updates on AI research and engineering, summarizing papers, repositories, benchmarks, safety, and systems in an accessible format. The platform covers a wide range of topics including agentic AI systems and safety validation, robotics and control, multimodal and speech models, applied drug discovery AI, federated/private ML, reasoning systems, optimization, and practical inference speedups. Each post provides clear explanations and links directly to the original research, such as arXiv, making complex AI advancements understandable for a broad audience. It aims to make research more accessible to professionals and students alike.

Amii (Alberta Machine Intelligence Institute) is a world-leading AI institute based in Canada, recognized as one of the country's AI centres of excellence. It operates on two primary fronts: advancing AI research through groundbreaking discoveries and transformative technologies, and facilitating the commercialization of AI by partnering with companies. Amii helps businesses, from startups to global enterprises, develop actionable AI strategies and custom AI solutions. The institute also focuses on inclusive AI, offering programs for literacy, training, and talent development, and is a global leader in Reinforcement Learning (RL).

DeepSeek-Prover-V1.5 is an open-source language model specifically designed for theorem proving within the Lean 4 proof assistant environment. It significantly enhances its predecessor, DeepSeek-Prover-V1, by optimizing both training and inference processes. The model is pre-trained on DeepSeekMath-Base and specializes in formal mathematical languages. It undergoes supervised fine-tuning with an advanced formal theorem proving dataset and further refinement through reinforcement learning from proof assistant feedback (RLPAF). A key innovation is RMaxTS, a Monte-Carlo tree search variant that employs an intrinsic-reward-driven exploration strategy to generate diverse proof paths. This approach has led to state-of-the-art results on benchmarks like miniF2F and ProofNet.

proudP is an FDA Class II uroflowmetry medical device that transforms a smartphone into an at-home urine flow test. Utilizing acoustic AI technology, it measures key uroflowmetry metrics such as maximum flow rate (Qmax), voided volume, and voiding time, and generates a flow rate graph. The app also includes the International Prostate Symptom Score (IPSS) test to help users monitor their prostate and bladder health. proudP offers a private, non-invasive way to track urinary symptoms, assess medication effectiveness, and generate reports for doctors, making it a valuable tool for individuals and healthcare providers alike.

FAR.AI is a non-profit organization dedicated to ensuring advanced AI is safe and beneficial for everyone through research and education. They explore a portfolio of high-potential research agendas, including robustness, interpretability, model evaluation, and alignment. FAR.AI also hosts and delivers a suite of events, such as alignment workshops and specialized conferences, bringing together global leaders in AI academia and industry. Additionally, their programs, like FAR.Labs and grantmaking initiatives, support innovation in trustworthy and secure AI by equipping researchers and organizations with necessary tools, resources, and connections.

EmergentX Digital Insights serves as a leading platform for institutional-grade insights and a professional community focused on digital assets and AI. It aims to accelerate the adoption of these technologies through independent insights, a robust expert network, and strategic events. The platform offers access to thought leaders via InfinityConnect (coming soon), facilitates connections among professionals, and provides advisory services for navigating the complexities of digital assets, AI, and blockchain. EmergentX also hosts world-class events and offers thought leadership content, empowering capital allocators, investors, and enterprises with actionable insights and opportunities for collaboration.

Research Locker is a comprehensive online platform designed to streamline research management for students, researchers, and professionals. It allows users to effortlessly save articles and organize PDFs within a clean, centralized workspace. A key feature is its ability to generate AI summaries, helping users quickly grasp the core content of their research materials. The tool also supports efficient knowledge management by enabling users to collect, store, and auto-tag content, ensuring that all research materials are easily accessible and manageable. This makes it an invaluable asset for anyone looking to enhance their research productivity and organization.

PrecisionLife leverages world-leading AI and genetic analysis to drive precision medicine beyond oncology, focusing on complex chronic diseases. The platform identifies causal disease biology and stratifies patients with mechanistic biomarkers, enabling the prediction and prevention of chronic conditions. It offers insights for improving health outcomes by predicting lifetime disease risk, determining underlying causes, and personalizing treatment and prevention strategies. For drug development, PrecisionLife discovers novel targets, accelerates clinical trials, and matches patients to appropriate treatments, maximizing success and de-risking programs. Their unique combinatorial analytics generate significant biological signals from less data, providing actionable insights for diagnostics and therapies.

Womxn in Data Science hosts the annual Woven Conference, Canada's first conference for women in data and AI, attracting over 6,000 attendees since 2018. The conference features three dynamic streams: 'Bits and Bytes' for technical deep dives, 'Dayta Dahta Decisions' for business leaders, and an 'Executive Lounge' for leadership discussions. It offers opportunities to learn real strategies from experienced professionals, connect with peers, and grow careers through practical insights and community support. Attendees can participate in workshops, listen to keynote speakers, and network with industry leaders, fostering an inclusive environment for women and allies in AI.

The UK Open Multimodal AI Network (UKOMAIN) is a national initiative funded by EPSRC, dedicated to fostering innovation and collaboration in the field of multimodal AI. Its core mission is to connect researchers, practitioners, and policymakers to facilitate the integration and analysis of diverse data types, including text, images, and sound. UKOMAIN actively supports various research projects, organizes workshops, and champions open science principles to address complex engineering research challenges within the multimodal AI domain. The network serves as a central hub for advancing the understanding and application of AI that can process and interpret information from multiple modalities simultaneously.

bsuite is an open-source collection of carefully-designed experiments aimed at investigating the core capabilities of reinforcement learning (RL) agents. It offers clear, informative, and scalable problems to address key issues in developing efficient and general learning algorithms. The suite automates the evaluation and analysis of agent behavior on these shared benchmarks, promoting reproducible and accessible research in RL. It includes various environments, configurable for different difficulty levels and random seeds, and provides tools for logging results and generating comprehensive analysis reports. bsuite supports integration with dm_env and OpenAI Gym interfaces, making it versatile for different RL frameworks. It also offers baseline agent implementations and facilitates running experiments on Google Cloud Platform.

ContrastiveSeg is an open-source project that implements the ICCV2021 (Oral) paper, "Exploring Cross-Image Pixel Contrast for Semantic Segmentation." This tool focuses on addressing the limitations of current semantic segmentation methods by incorporating "global" context, specifically rich semantic relations between pixels across different images. Inspired by unsupervised contrastive representation learning, ContrastiveSeg proposes a pixel-wise contrastive framework for fully supervised semantic segmentation. It enforces pixel embeddings of the same semantic class to be more similar than those from different classes, introducing a novel pixel-wise metric learning paradigm. The method can be integrated into existing segmentation frameworks without additional overhead during testing and has shown consistent performance improvements on datasets like Cityscapes, PASCALContext, and COCO-Stuff when used with models like DeepLabV3, HRNet, and various backbones.