RePaint uses a combination of 3D printing and deep learning to authentically recreate paintings, regardless of lighting conditions. The system was tested on oil paintings and found to be more than four times more accurate than state-of-the-art physical models.
Researchers identify potential causes of bias in machine learning systems and demonstrate how changing data collection methods can reduce bias without compromising accuracy. They suggest identifying clusters of patients with high disparities in accuracy to inform data collection decisions.
The MIT CSAIL team has developed a new approach called DAWG to securely partition cache memory, making it harder for hackers to exploit Meltdown and Spectre vulnerabilities. This method can have immediate applications in cloud computing, especially in fields like medicine and finance.
Researchers from MIT CSAIL developed a system that uses machine learning to determine if a news source is accurate or biased. The system achieved 65% accuracy in detecting factuality levels and 70% accuracy in detecting bias, using common linguistic features across the source's stories.
A new system, MoSculp, uses algorithmic processing of 2D videos to generate 3D printed motion sculptures, providing a detailed study of human movement for athletes, dancers, and others. The system has been shown to improve visualization and analysis of complex motion in over 75% of user studies.
Researchers at MIT CSAIL have developed a robot system that can inspect and understand objects, allowing it to perform specific tasks such as picking up objects from a cluttered space. The system, called Dense Object Nets, uses a self-supervised approach and doesn't require human annotations.
ReMix uses wireless signals to pinpoint the location of ingestible implants inside the body with centimeter-level accuracy. This technology has the potential to improve proton therapy cancer treatment by allowing doctors to track tumors in real-time and deliver targeted radiation.
A new cryptographic system, AUDIT, is proposed to improve government surveillance accountability while maintaining confidentiality. The system uses a public ledger and cryptographic commitments to ensure transparency and accountability.
Researchers at MIT's CSAIL have developed a system that allows people to control robots using brainwaves and hand gestures. The system uses EEG and EMG feedback to detect errors and correct robot mistakes on multi-choice problems, enabling more natural human-robot interactions.
A team from MIT's CSAIL has developed a system that uses artificial intelligence to analyze radio signals bouncing off people's bodies, allowing for dynamic stick figures to be created of their postures and movements. This technology could monitor diseases like Parkinson's and MS, as well as provide security for elderly people.
Researchers at MIT's CSAIL have developed a system called VirtualHome that can simulate detailed household tasks and train artificial agents to execute them. The system uses natural language descriptions of activities, such as making coffee or setting the table, to instruct robots on how to complete tasks.
Researchers at MIT CSAIL have developed a new framework called MapLite that allows self-driving cars to navigate on unpaved country roads without detailed maps. The system combines GPS data with sensors to detect road conditions and drive autonomously, with promising results in testing.
Researchers at MIT's CSAIL have developed RoadTracer, an automated method to build road maps that is 45 percent more accurate than existing approaches. The system uses data from aerial images and creates maps step-by-step, tracing out roads one step at a time.
A new tool called Squadbox, developed by MIT CSAIL, enables people who have been targeted by online harassment to coordinate a squad of friends to filter messages and shield themselves from abusive content. The 'friend-sourcing' approach aims to augment human support with tools in a meaningful way.
AutoSaw, a system developed by MIT CSAIL, lets users design and customize furniture templates using expert knowledge and robotics for cutting tasks. The system uses small mobile robots to cut lumber with minimal risk of injury.
NanoMap enables drones to fly through cluttered spaces like cities at high speeds using a simple yet effective approach that considers the drone's position in the world over time. The system reduces crash rates from over 90% to just 2%, making it suitable for applications such as search-and-rescue, defense, and package delivery.
Researchers from MIT's CSAIL present ColorFab, a method for repeatedly changing the colors of 3-D printed objects after fabrication using custom ink and UV light. The system can recolor a multi-colored object in under 20 minutes, with the goal of reducing waste and increasing customization.
Researchers from MIT's CSAIL team show that keeping an equal distance between cars reduces traffic jams, enabling drivers to reach their destinations almost twice as quickly. The approach, known as bilateral control, could be achieved through software and hardware updates, without requiring extensive infrastructure changes.
The CSAIL team developed a VR system that lets users teleoperate robots using an Oculus Rift headset. The system mimics the user's movements to complete various tasks, making it feel like they are inside the robot's head. This technology could enable blue-collar workers to telecommute and benefit from the IT revolution.
Researchers at MIT's CSAIL developed ComText, a system that allows robots to understand contextual commands, enabling them to perform complex tasks. With this technology, Baxter, a two-armed humanoid robot, achieved success in executing commands with a 90% accuracy rate.
Researchers at MIT CSAIL developed an interactive design system called Interactive Robogami that allows users to design and 3D-print custom robots in minutes. The system uses simulations, algorithms, and interactive feedback to ensure feasible designs and guarantee stability and speed.
Researchers developed a machine-learning approach called ICU Intervene that takes large amounts of ICU data to determine treatments for different symptoms. The system uses deep learning to make real-time predictions, learning from past cases to suggest critical care interventions.
A new AI system developed by MIT researchers reduces online video rebuffering and improves quality, adapting to different network conditions. The system achieves higher-quality streaming with less rebuffering than existing approaches.
Researchers from MIT's CSAIL have developed InstantCAD, a tool that enables designers to interactively edit and optimize CAD models using a streamlined workflow. This allows for real-time improvements and optimizations, reducing the time required to test and refine complex designs.
Researchers from MIT's CSAIL developed an AI system called Pic2Recipe that can analyze photos of food and predict ingredients and suggest similar recipes. The system was trained on a database of over 1 million recipes and showed impressive results, particularly with desserts like cookies or muffins.
Researchers from MIT's CSAIL have developed a new system called Home3D that allows users to watch 3D movies at home without glasses. The system converts traditional 3D movies into a format compatible with automultiscopic displays, which show multiple images simultaneously.
Researchers at MIT's CSAIL have developed a system of quadcopter drones that can both fly and drive through urban environments with ease. The drones, equipped with wheels on the bottom, can navigate around obstacles in both air and ground modes, making them ideal for transporting objects or rescuing people in disaster zones.
A new MIT-developed device can accurately measure walking speed using wireless signals, providing insights into cognitive decline, cardiac disease, and other health issues. The system has 95-99% accuracy and can be placed on a person's home wall without intruding on daily life.
Researchers from MIT CSAIL create an app called WaitSuite that tests users on vocabulary words during wait times, such as WiFi connectivity and email updates. The system, called "wait-learning," helps users make the most of small moments, enabling them to focus on primary tasks and learn new words simultaneously.
A new wearable AI system developed by MIT researchers can predict the tone of conversations with high accuracy, analyzing audio and vital-sign data. The system has shown promise in understanding social interactions, particularly for people with anxiety or Asperger's syndromes.
Researchers developed a machine learning system that detects differences in accelerometer data between individuals with muscle tension dysphonia and controls. The system improved after receiving voice therapy, suggesting potential for wearable devices providing real-time feedback.
Researchers from MIT's CSAIL have developed an imaging technique called Interactive Dynamic Video (IDV) that lets users reach in and interact with objects in videos. By analyzing video clips for 'vibration modes,' the team can predict how objects will move in new situations, enabling realistic simulations.
A new display technology, Cinema 3D, allows viewers to watch 3-D movies without glasses, using a special array of lenses and mirrors. The system enables high-resolution images to be replicated across the theater, offering an immersive cinematic experience from any seat.
Researchers at MIT CSAIL discovered brain organizes objects based on size, with distinct regions for big and small objects. This finding has major implications for fields like robotics and could lead to better understanding of brain organization and mapping information.
The MIT CSAIL project aims to develop a desktop technology that enables personalized robot design, customization, and printing within hours. Researchers hope to create an end-to-end process for building physical machines using simple printing processes.