Dynamic scene reconstruction is a long-term challenge in the field of 3D vision. Recently, the emergence of 3D Gaussian Splatting has provided new insights into this problem. Although subsequent efforts rapidly extend static 3D Gaussian to dynamic scenes, they often lack explicit constraints on object motion, leading to optimization difficulties and performance degradation. To address the above issues, we propose a novel deformable 3D Gaussian splatting framework called MotionGS, which explores explicit motion priors to guide the deformation of 3D Gaussians. Specifically, we first introduce an optical flow decoupling module that decouples optical flow into camera flow and motion flow, corresponding to camera movement and object motion respectively. Then the motion flow can effectively constrain the deformation of 3D Gaussians, thus simulating the motion of dynamic objects. Additionally, a camera pose refinement module is proposed to alternately optimize 3D Gaussians and camera poses, mitigating the impact of inaccurate camera poses. Extensive experiments in the monocular dynamic scenes validate that MotionGS surpasses state-of-the-art methods and exhibits significant superiority in both qualitative and quantitative results.

This paper focuses on detecting clickbait posts on the Web. These posts often use eye-catching disinformation in mixed modalities to mislead users to click for profit. That affects the user experience and thus would be blocked by content provider. To escape detection, malicious creators use tricks to add some irrelevant non-bait content into bait posts, dressing them up as legal to fool the detector. This content often has biased relations with non-bait labels, yet traditional detectors tend to make predictions based on simple co-occurrence rather than grasping inherent factors that lead to malicious behavior. This spurious bias would easily cause misjudgments. To address this problem, we propose a new debiased method based on causal inference. We first employ a set of features in multiple modalities to characterize the posts. Considering these features are often mixed up with unknown biases, we then disentangle three kinds of latent factors from them, including the invariant factor that indicates intrinsic bait intention; the causal factor which reflects deceptive patterns in a certain scenario, and non-causal noise. By eliminating the noise that causes bias, we can use invariant and causal factors to build a robust model with good generalization ability. Experiments on three popular datasets show the effectiveness of our approach.

We present a practical distillation approach to fine-tune LLMs for invoking tools in real-time applications. We focus on visual editing tasks; specifically, we modify images and videos by interpreting user stylistic requests, specified in natural language ("golden hour"), using an LLM to select the appropriate tools and their parameters to achieve the desired visual effect. We found that proprietary LLMs such as GPT-3.5-Turbo show potential in this task, but their high cost and latency make them unsuitable for real-time applications. In our approach, we fine-tune a (smaller) student LLM with guidance from a (larger) teacher LLM and behavioral signals. We introduce offline metrics to evaluate student LLMs. Both online and offline experiments show that our student models manage to match the performance of our teacher model (GPT-3.5-Turbo), significantly reducing costs and latency. Lastly, we show that fine-tuning was improved by 25% in low-data regimes using augmentation.

Recent progress in natural language generation has raised dual-use concerns. While applications like summarization and translation are positive, the underlying technology also might enable adversaries to generate neural fake news: targeted propaganda that closely mimics the style of real news. Modern computer security relies on careful threat modeling: identifying potential threats and vulnerabilities from an adversary's point of view, and exploring potential mitigations to these threats. Likewise, developing robust defenses against neural fake news requires us first to carefully investigate and characterize the risks of these models. We thus present a model for controllable text generation called Grover.

Alex Galvagni, CEO of Age of Learning and a former artificial intelligence researcher with NASA, says advances in AI now make it possible to deliver to children "a personalized and supportive" experience in education. This photo combo shows the 2024 Nobel Prize winners in Physics, professor John Hopfield, left, of Princeton University, and professor Geoffrey Hinton, of the University of Toronto, on Tuesday, Oct. 8, 2024. FOUNDATIONAL WORK: Two pioneers of artificial intelligence -- John Hopfield and Geoffrey Hinton -- won the Nobel Prize in physics Tuesday for helping to create the building blocks of machine learning that are revolutionizing the way we work and live, but also create new threats for humanity. UBER EV: Ride-sharing platform Uber on Tuesday announced that the company is taking new steps to promote the use of electric vehicles (EVs) on its platform. 'BETTER JOB': The United Nations (U.N.) advisory body on artificial intelligence (AI) last week issued seven recommendations to address AI-related risks, but an expert told Fox News Digital the points do not cover critical areas of concern.

Back in the old days, there was no sure-fire indicator of box office poison more than a video game adaptation. That has changed in recent years and now all kinds of gaming mascots are getting their chance to appear in a major motion picture or, at the very least, a streaming series. They're now making a movie based on Shinobi, as reported by Deadline. For the uninitiated, Shinobi is a famous hack-and-slash game developed by Sega in which you play as a ninja. There have been plenty of sequels throughout the years, though they mostly share the same basic story.

Structured distributions, i.e. distributions over combinatorial spaces, are commonly used to learn latent probabilistic representations from observed data. However, scaling these models is bottlenecked by the high computational and memory complexity with respect to the size of the latent representations. Common models such as Hidden Markov Models (HMMs) and Probabilistic Context-Free Grammars (PCFGs) require time and space quadratic and cubic in the number of hidden states respectively. This work demonstrates a simple approach to reduce the computational and memory complexity of a large class of structured models. We show that by viewing the central inference step as a matrix-vector product and using a low-rank constraint, we can trade off model expressivity and speed via the rank. Experiments with neural parameterized structured models for language modeling, polyphonic music modeling, unsupervised grammar induction, and video modeling show that our approach matches the accuracy of standard models at large state spaces while providing practical speedups.

Amazon's fall Prime Day sale--also known as Big Deals Days--ends tonight. It's October, yes, but it's never too early to jump on that holiday gift shopping. We've combed through the deals and found the best ones, based on our years of testing and reviewing. WIRED's picks for the best Prime Day deals only include products someone from our team has personally tested and reviewed. We track prices using several tools to avoid falling for fake discounts. There are no shoddy knockoffs or overpriced products among our recommendations, just good deals on good stuff. We've linked our reviews and buying guide throughout to help you make fully informed buying decisions. We test products year-round and handpicked these Prime Day deals. We'll update this guide regularly throughout Prime Day by adding fresh deals and removing dead deals. This is our favorite e-reader. You'll have the choice between the base Paperwhite and the Signature Edition (8/10, WIRED Recommends), which comes with 16 gigabytes ...

On this week's show, Dana and Stephen are joined by Supreme Friend of the Podcast (SFOP) Isaac Butler, author of The Method: How the Twentieth Century Learned to Act. The trio first explores The Wild Robot, DreamWork Animation's handcrafted, lovingly made film that's the surprise of the year. Lupita Nyong'o voices ROZ, an old-fashioned robot powered by supremely advanced A.I. who must learn about and adapt to her new wild surroundings. Then, they dissect Nobody Wants This, a new Netflix series starring Kristen Bell (who plays a sex podcaster) and Adam Brody as a hot rabbi. Although there are obvious charms, the show's "will they, won't they" rom-com beats can often feel, at best, gratingly familiar, and at worst, bizarre and unthoughtful, particularly in its portrayal of Jewish women.

If you're looking to upgrade your entertainment setup, finding the best projector could be the perfect solution. Whether you're into binge-watching shows, hosting outdoor movie nights or even leveling up your gaming experience, modern projectors can help you do it all. Some are fantastic for creating that full home-theater vibe, while others are so good they could even replace your TV, offering huge screen sizes, sharp image quality and built-in smart features. Many projectors are portable enough to take outside, making them great for BBQs, yard parties, or just enjoying a cozy movie night under the stars. Some are even designed for easy room-to-room transport, meaning you can switch up your viewing experience wherever you are. If you're thinking of stepping up your viewing game, we've tested some of the best projectors out there to help you find the right one for your needs. As mentioned, ultra-short-throw models have rapidly established themselves in the market due to the extra performance and convenience, and all manufacturers sell at least a couple of models. Within the ultra-short-throw category, We'll compare two price categories: under 7,000 and 3,500, with three projectors each.