University Research

The evolution of machine learning has brought significant advancements in language models, which are foundational to tasks like text generation and question-answering. Among these,...

The Allen Institute for AI (AI2) has announced the release of Tülu 3, a state-of-the-art family of instruction-following models designed to set a new...

Large language models (LLMs) have revolutionized various applications across industries by providing advanced natural language processing capabilities. These models' ability to generate, understand, and...

Deep features are pivotal in computer vision studies, unlocking image semantics and empowering researchers to tackle various tasks, even in scenarios with minimal data....

In areas with clearly defined reward functions, like games, reinforcement learning (RL) has outperformed human performance. Unfortunately, it is difficult or impossible for many...

Despite recent advances in RL research, the ability to generalize to new tasks remains one of the major issues in both reinforcement learning (RL)...

Image and video editing are two of the most popular applications for computer users. With the advent of Machine Learning (ML) and Deep Learning...

Artificial intelligence is making noteworthy strides in the field of computer vision. One key area of development is deep learning, where neural networks are...

Finding all of the "objects" in a given image is the groundwork of computer vision. By creating a vocabulary of categories and training a...

Latent diffusion models (LDMs), a subclass of denoising diffusion models, have recently acquired prominence because they make generating images with high fidelity, diversity, and...

Stock market behavior forecasting is a crucial undertaking that requires careful attention since, with the right choices, a successful prediction of stock prices might...

Extreme weather conditions have become a typical occurrence, especially in recent years. Climate change is the main factor to blame for such extreme weather-related...

Though recent advances have been made in large language model (LLM) reasoning, LLMs still have a hard time with hierarchical multi-step reasoning tasks like...

When it comes to machine learning (ML) and artificial intelligence (AI), having a good quality dataset with sufficient data points is of fundamental importance...

In many real-world computer vision problems, such as healthcare, labeled training data can be scarce, leading to the development of machine learning models that...

By improving the in-context learning quality of smaller and open source models, more researchers and organizations can study and apply the technology. One exciting...

Topping the list of the world's worst malignancies is lung cancer, which is expected to claim 1.7 million lives around the globe in 2020....

In recent years, the possible applications of text-to-image models have increased enormously. However, image editing to human-written instruction is one subfield that still has...

Another day and another blog post about diffusion models. Diffusion models were probably one of the hottest, if not the hottest, topics in the...

Until recently, the results of AI models were kept secret. Even if researchers know the dataset a model was trained on, they may still...

Videos with six degrees of freedom (6-DoF) let viewers freely explore an area by allowing them to adjust their head position (3 degrees of...

Capturing and synthesizing realistic human motion trajectories can be extremely useful in virtual reality, game character animations, CGI, and robotics. We need large datasets...

Dimensionality reduction (DR) is a method for analyzing high-dimensional data that involves minimizing the number of variables taken into account. Data visualization in two...

Humans find some of the comprehension difficult to understand. There is a limitation to the ability to understand some sentences, and find it difficult...

In surroundings humans have created, line segments are common and efficiently convey the underlying picture structure. They complement feature points nicely because of their...

Program comprehension is the process through which software engineers use the source code as their main source of information to comprehend the behavior of...

A quick rollout of mRNA-based vaccinations against severe acute respiratory syndrome coronavirus 2 demonstrates the enormous promise of mRNA-based therapeutics as a modular therapeutic...

The capacity of Neural Radiance Fields (NeRF) and its derivatives to accurately recreate real-world 3D scenes from 2D photos and allow high-quality, photorealistic new...

It is crucial to produce high-fidelity digital humans for numerous applications, including immersive telepresence, AR/VR, 3D graphics, and the developing metaverse. Getting customized avatars...

Although their presence is not as significant as projected by Sci-Fi movies from the 90s, robots are becoming essential in our daily lives with...

Language models are transforming science and society thanks to recent technological advances in artificial intelligence and machine learning. However, language models are still somewhat...

The development of large language models (LLMs) has significantly advanced artificial intelligence (AI). With the use of learned knowledge, these models have astounding possibilities...

Over the last decade, computational catalysis has emerged as one of the most active research areas, and it is currently a vital tool for...

The tear film must spread quickly and uniformly throughout the ocular surface for clear vision and good ocular health. One of the ocular morbidities...

We have all encountered new words we were unfamiliar with while reading scientific research papers. It can be difficult for a total novice to...

There are many multi-modal foundation models existing, which are typically trained on millions of natural images and their text captions. Medical images are just...

The two of the biggest problems for both large and small enterprises are analysis and storage. To begin, the rate at which Big Data...

Estimating the trajectory of items in a video clip is the goal of multiple object tracking (MOT). While specific MOT standards only consider tracking...

Virtual and augmented reality (AR/VR) systems have gained a lot of attention in the last ten years. The main aim being making user experiences...

Even with the most recent technological advances, AI has yet to be able to produce conclusions that are completely impartial or 100% ethically correct....

Graphs, an abstract, non-linear data type, are frequently employed to illustrate links between different types of data, such as social media connections, hierarchies, financial...

Since the last few years, the clinical NLP community has worked incredibly hard to extract crucial variables hidden in digital records. This idea of...

Applying neural scaling laws to machine learning models, which means increasing the number of computations, the size of the model, and the number of...

A form of social learning, imitation is how new behaviors are picked up. Understanding how to communicate, interact socially, and control one's emotions while...

A particular human tendency is a desire to comprehend the complex world around them and to communicate that understanding to others. This is why...

Machine Learning (ML), or more precisely, Deep Learning (DL), has revolutionized the field of Artificial Intelligence (AI) and made tremendous breakthroughs in numerous areas,...

Caltech research introduces 'SNIascore,' a method for spectroscopically classifying thermonuclear supernovae (SNe Ia) based on very low-resolution (R 100) data based on deep learning....

Reinforcement learning is a machine learning method in which an artificial agent learns from its mistakes. The agent receives a reward from the researchers...

Many cultures place a high value on dance as a means of expression, communication, and social connection. However, producing new dances or dance animations...

Benchmarks orient AI. They encapsulate ideals and priorities that describe how the AI community should progress. When properly developed and analyzed, they allow the...

The methods currently used to correct systematic issues in NLP models are either fragile or time-consuming and prone to shortcuts. Humans, on the other...

Cell types, basement membranes, and connective structures that organize tissues and tumors can be found in length ranges ranging from microscopic organelles to whole...

Artificial Intelligence (AI) field has been busy with handling the burst in generative models for the last couple of months. The open-source release of...

Continuous-time neural networks are one subset of machine learning systems capable of taking on representation learning for spatiotemporal decision-making tasks. Continuous differential equations are...

Although large language models (LLMs) have shown astounding performance on tasks involving interactive decision-making and language understanding, their capacities for reasoning (such as chain-of-thought...

Generative Adversarial Networks (or just GANs) have been widely used to generate synthetic data for different applications in recent years. The most commonly considered...

Mathematical reasoning is a fundamental requirement for general-purpose artificial intelligence systems. These tasks can have a wide range of complexity, and they might be...

Where are the facts kept in a Large language model or LLM? For two reasons, we are curious about how and where a model keeps...

The first extensive text-to-image prompt dataset is called DiffusionDB. It has 2 million Stable Diffusion-generated photos that were produced using prompts and hyperparameters provided...

Recent progress in generative models has paved the way to a manifold of tasks that some years ago were only imaginable. With the help...

Our earth is gorgeous, with majestic mountains, breathtaking seascapes, and tranquil forests. Flying past intricately detailed, three-dimensional landscapes, picture yourself taking in this splendor...

The development of neuromorphic electronics depends on the effective mimic of neurons. But artificial neurons aren't capable of operating in biological environments. Organic artificial...

Large models have improved performance on a wide range of modern computer vision and, in particular, natural language processing problems. However, issuing patches to...

Action recognition, or teaching a machine to recognize human actions, has a wide range of potential applications. For example, it may be used to...

Augmented reality (AR) is on its way to becoming a part of our daily lives. We can define it as placing a virtual object...

Our environment is filled with acoustic information. When we walk into a hall, the acoustic and visual information can help us map the hall...

Topological materials are a special kind of material that have different functional properties on their surfaces than on their interiors. One of these properties...

Nowadays, it is difficult to pass a day without reading/hearing about a new application of diffusion models if you are following the news about...

Researchers have developed a new technique for performing computations straight on smart home gadgets. Their method moves the memory-intensive machine-learning model operations to a...

Art is a fascinating yet extremely complex discipline. Indeed, the creation of artistic images is often not only a time-consuming problem but also requires...

High-resolution picture synthesis using denoising diffusion probabilistic models (DDPMs) with classifier-free guidings, such as DALLE 2, GLIDE, and Imagen, has reached state-of-the-art results. The...

Visual data modeling, such as photographs and videos, is a canonical problem in deep learning. Many current deep learning backbones with good performance on...

Deep learning methods have been a game-changer in lots of applications. They have been the core components of field advancements in the last decades,...

Information is processed by microprocessors in computers, data centers, and smartphones by manipulating electrons that pass through solid semiconductors. However, our brains have a...

The ground-breaking research conducted in the last ten years is mainly responsible for the remarkable achievements gained in machine learning. Machine learning applications are...

We all saw the importance of good-quality video conferencing tools during COVID lockdowns. Education, entertainment, work meetings, and family visits became video conferences, and...

By prompting a text-to-image model we can generate images of a wide variety of objects. With clever prompting, it’s also possible to synthesize different...

In this article, UC Berkeley and Meta researchers demonstrate how an adaptive controller can be trained to rotate various objects over the z-axis using...

The goal of the content-based model search is introduced, which tries to locate the most relevant deep image generative models that fulfill a user's...

Neural Radiance Fields (NeRFs) is a revolutionary approach for 3D representation that uses a multi-layer perceptron to describe the geometry and view-dependent appearance of...

The necessity for accurate and economical gigapixel image analysis has risen as whole-slide imaging has become more widely used. Deep learning is at the...

In recent years, CNN architectures have made remarkable progress in various challenging vision tasks, including semantic segmentation and object classification. Large activation maps in...

Molecular docking is essential in computational drug design since it allows one to predict the location, orientation, and conformation of a ligand when attached...

Proteins have been intensively explored as a therapeutic medium due to their relevance, and they comprise a fast-rising proportion of authorized medicines. Proteins are...

Studies have shown that it is possible to adjust the pre-trained model for newly collected sensory data after deployment for on-device training. The model...

The production of human motion is a critical job in computer animation, with applications ranging from gaming to robots. It is a tricky topic...

Synthetic Data (SD) Metrics is a new tool developed by DataCebo, a startup born out of MIT's Computer Science & Artificial Intelligence Laboratory (CSAIL)...

For various applications, including question answering, textual entailment, and summarization, large pre-trained language models have demonstrated excellent text production capabilities. However, most work using...

The researchers discovered that by looking into how Apple used the local differential privacy (LDP) model, they could determine people's favorite emoji skin tone...

While today’s deep neural networks (DNNs) drive AI’s deep-learning revolution, determining a DNN’s appropriate complexity remains challenging. If a DNN is too shallow, its...

Amyotrophic lateral sclerosis (ALS), Alzheimer's disease, and Parkinson's disease are all examples of neurodegenerative disorders, which are complex and chronic illnesses. These disorders have...

Deep generative models are a popular data generation strategy used to generate high-quality samples in pictures, text, and audio and improve semi-supervised learning, domain...

Artistic portraits are everywhere in their everyday lives and in the creative sectors that inform the arts, social media avatars, movies, entertainment advertising, etc....

Deep learning's success is due to its capacity to tackle some enormous non-convex optimization problems with relative simplicity. Even though non-convex optimization is NP-hard,...

As Machine Learning becomes increasingly prevalent in software, a new subfield known as MLOps (short for ML Operations) has evolved to organize the "collection...

Voice-to-text processing has advanced significantly in recent years, making the occasional failures in AI-powered speech recognition systems little more than curious outliers. However, most...

In 1947, the genius mathematician and computer scientist Alan Turing anticipated the current state of machine learning research by stating that a machine should...

The challenge of motion estimation is important to computer vision and has far-reaching implications. Tracking makes it possible to create models of an object's...

Nowadays, several social networking sites heavily rely on connection recommendations. Surveys reveal that connection suggestions may make up more than 50% of the social...

Most AI models in medical machine learning need labeled or annotated datasets for training since it teaches the models to recognize diseases accurately. Since...