Photonic Processor Could Enable Ultrafast AI Computations with Extreme Energy Efficiency

Adam Zewe | MIT News

This new device uses light to perform the key operations of a deep neural network on a chip, opening the door to high-speed processors that can learn in real-time.

Graphic showing light emanating from a cubic crystal and passing through a material with an array of square holes. A lattice of atoms appears on the other side

AI Method Radically Speeds Predictions of Materials’ Thermal Properties

Adam Zewe | MIT News

The approach could help engineers design more efficient energy-conversion systems and faster microelectronic devices, reducing waste heat.

Cartoon with several online chat windows saying “Oops something went wrong,” and one in the center with text bubbles showing it is continuing to perform.

A New Way to Let AI Chatbots Converse All Day without Crashing

Adam Zewe | MIT News

Researchers developed a simple yet effective solution for a puzzling problem that can worsen the performance of large language models such as ChatGPT.

A cell phone peeks out from the pocket of a person wearing jeans, a belt, and a plaid shirt. In the background and on the cell phone’s screen are stylized connected nodes representing a neural network.

Technique Enables AI on Edge Devices to Keep Learning Over Time

Adam Zewe | MIT News

With the PockEngine training method, machine-learning models can efficiently and continuously learn from user data on edge devices like smartphones.

Memristive Arrays Analog Computing and Neuromorphic Engineering

Memristive Crossbar Arrays for Analog In-Memory Computing and Neuromorphic Engineering

Wednesday, October 11, 2023 | 12:00 - 1:00pm ET
Hybrid

Grier A (34-401A)
50 Vassar Street Cambridge, MA

Rendering shows a figure standing in castle ruins, and a wooden box in foreground.

Helping Computer Vision and Language Models Understand What They See

Adam Zewe | MIT News

Researchers use synthetic data to improve a model’s ability to grasp conceptual information, which could enhance automatic captioning and question-answering systems.

Building Tools To Learn Human Brain Processes

Imagination + AI, MIT CSAIL, Forbes

The CSAIL Imagination in Action @ MIT Symposium aimed to educate and motivate entrepreneurs on building successful AI-focused companies, attracting a diverse audience passionate about AI's transformative potential.

Realistic drawing of a printed circuit board, with circuit lines projecting above it in the shape of a brain

Machine-learning System Based on Light Could Yield More Powerful, Efficient Large Language Models

Elizabeth A. Thomson | Materials Research Laboratory

MIT system demonstrates greater than 100-fold improvement in energy efficiency and a 25-fold improvement in compute density compared with current systems.

Computer motherboard image composed of green and orange colored rays

Busy GPUs: Sampling and Pipelining Method Speeds Up Deep Learning on Large Graphs

Lauren Hinkel | MIT-IBM Watson AI Lab

New technique significantly reduces training and inference time on extensive datasets to keep pace with fast-moving data in finance, social networks, and fraud detection in cryptocurrency.