MAROKO133 Update ai: Adobe Research Unlocking Long-Term Memory in Video World Models with

📌 MAROKO133 Eksklusif ai: Adobe Research Unlocking Long-Term Memory in Video World

Video world models, which predict future frames conditioned on actions, hold immense promise for artificial intelligence, enabling agents to plan and reason in dynamic environments. Recent advancements, particularly with video diffusion models, have shown impressive capabilities in generating realistic future sequences. However, a significant bottleneck remains: maintaining long-term memory. Current models struggle to remember events and states from far in the past due to the high computational cost associated with processing extended sequences using traditional attention layers. This limits their ability to perform complex tasks requiring sustained understanding of a scene.

A new paper, “Long-Context State-Space Video World Models” by researchers from Stanford University, Princeton University, and Adobe Research, proposes an innovative solution to this challenge. They introduce a novel architecture that leverages State-Space Models (SSMs) to extend temporal memory without sacrificing computational efficiency.

The core problem lies in the quadratic computational complexity of attention mechanisms with respect to sequence length. As the video context grows, the resources required for attention layers explode, making long-term memory impractical for real-world applications. This means that after a certain number of frames, the model effectively “forgets” earlier events, hindering its performance on tasks that demand long-range coherence or reasoning over extended periods.

The authors’ key insight is to leverage the inherent strengths of State-Space Models (SSMs) for causal sequence modeling. Unlike previous attempts that retrofitted SSMs for non-causal vision tasks, this work fully exploits their advantages in processing sequences efficiently.

The proposed Long-Context State-Space Video World Model (LSSVWM) incorporates several crucial design choices:

1. Block-wise SSM Scanning Scheme: This is central to their design. Instead of processing the entire video sequence with a single SSM scan, they employ a block-wise scheme. This strategically trades off some spatial consistency (within a block) for significantly extended temporal memory. By breaking down the long sequence into manageable blocks, they can maintain a compressed “state” that carries information across blocks, effectively extending the model’s memory horizon.
2. Dense Local Attention: To compensate for the potential loss of spatial coherence introduced by the block-wise SSM scanning, the model incorporates dense local attention. This ensures that consecutive frames within and across blocks maintain strong relationships, preserving the fine-grained details and consistency necessary for realistic video generation. This dual approach of global (SSM) and local (attention) processing allows them to achieve both long-term memory and local fidelity.

The paper also introduces two key training strategies to further improve long-context performance:

• Diffusion Forcing: This technique encourages the model to generate frames conditioned on a prefix of the input, effectively forcing it to learn to maintain consistency over longer durations. By sometimes not sampling a prefix and keeping all tokens noised, the training becomes equivalent to diffusion forcing, which is highlighted as a special case of long-context training where the prefix length is zero. This pushes the model to generate coherent sequences even from minimal initial context.
• Frame Local Attention: For faster training and sampling, the authors implemented a “frame local attention” mechanism. This utilizes FlexAttention to achieve significant speedups compared to a fully causal mask. By grouping frames into chunks (e.g., chunks of 5 with a frame window size of 10), frames within a chunk maintain bidirectionality while also attending to frames in the previous chunk. This allows for an effective receptive field while optimizing computational load.

The researchers evaluated their LSSVWM on challenging datasets, including Memory Maze and Minecraft, which are specifically designed to test long-term memory capabilities through spatial retrieval and reasoning tasks.

The experiments demonstrate that their approach substantially surpasses baselines in preserving long-range memory. Qualitative results, as shown in supplementary figures (e.g., S1, S2, S3), illustrate that LSSVWM can generate more coherent and accurate sequences over extended periods compared to models relying solely on causal attention or even Mamba2 without frame local attention. For instance, on reasoning tasks for the maze dataset, their model maintains better consistency and accuracy over long horizons. Similarly, for retrieval tasks, LSSVWM shows improved ability to recall and utilize information from distant past frames. Crucially, these improvements are achieved while maintaining practical inference speeds, making the models suitable for interactive applications.

The Paper Long-Context State-Space Video World Models is on arXiv

The post Adobe Research Unlocking Long-Term Memory in Video World Models with State-Space Models first appeared on Synced.

🔗 Sumber: syncedreview.com

📌 MAROKO133 Breaking ai: Faces of hundreds of years old Colombian mummies reconstr

One of the most reputable facial reconstruction labs brought the faces behind four funerary masks back to vivid life, revealing the extraordinary craftsmanship of Eastern Colombia between the 13th and 17th centuries.

For a new study, archaeologists from Face Lab at Liverpool John Moore University unmasked four mummies from the Colombian Institute of Anthropology and History. The way pre-Colombian cultures of South America crafted these masks allowed them to fuse with the remains, so that “the bodies seemed to be alive,” according to The Independent.

With their faces so permanently and impressively preserved within the interior of the funerary masks, researchers were able to reconstruct the damaged masks without physically removing them, breathing new life into the unreachable past.

A researcher from Face Lab told The Independent that the project “highlights the cultural practices of the indigenous peoples of South America,” and will hopefully draw more interest towards these “incredible civilizations.”

Masks crafted on the faces of the dead

Firstly, the masks are extraordinary examples of craftsmanship, “the one ones known to exist in Colombia,” The Independent continued, due to their ability to fuse with the face, leaving a permanent imprint that survived the test of time.

Ancient Origins reports that the masks were discovered in graves that had been looted, so they couldn’t provide any context beyond who they belonged to.

The four masks, respectively, sat on the faces of a 6 to 7-year-old child, a female in her 60s, and two young adult males, as per Live Science. These stylized masks were made of resin, clay, and maize with touches of gold and beads around the eyes. They were found in the Eastern Cordillera, a region in the Colombian Andes, and are thought to date between 1216 and 1797 AD.

According to Ancient Origins, they were formed directly on the anterior of the skull, and covered the entire face and jaw. Given that they were crafted on their faces, it attests to the skill that the funerary artist possessed.

Technology achieves the impossible

The Face Lab team digitally “peeled back” the masks to reveal the face and used CT X-ray scans to capture 3D images of it.

The Independent explained that they took 2D images of each sample and then analyzed the scans to reconstruct their faces. A software enabled them to add facial tissue, and “a haptic touch stylus pen,” Live Science said, allowed them to “virtually sculpt” muscles onto the digital skulls.

Ancient Origins added that they went so far as to utilize facial measurements of Colombian men living today, which they couldn’t do for the female or child, so they had to use their own expertise in anatomy, which the craftsmen did, to fill in the missing details.

“We used a digital sculpting process to rebuild each face. With the help of a stylus and specialized software, we digitally recreated muscle, skin, and fat onto the skulls.” Dr. Jessica Liu, project manager at Face Lab told Ancient Origins.

A sophisticated culture

Beyond the significance of recent advancements in technology that have enabled facial reconstructions to make headlines lately, researchers admitted to Ancient Origins that they are limited in their ability to capture wrinkles, freckles, and pores. Their skin tone, hair, and eye color; these images shouldn’t be taken as 100% true, but rather a meeting of fact and interpretation.

However, the recent project from Face Lab brings a remarkable piece of history to light, combining technology, craftsmanship, and funerary practices to reveal how sophisticated these cultures’ understandings of human anatomy were, also hinting at their beliefs around death. They held “elaborate burial practices” that attempted to preserve their dead for modern viewers to remember in exquisite detail, Ancient Origins concludes.

🔗 Sumber: interestingengineering.com