📌 TOPINDIATOURS Hot ai: Chinese scientists expand range of quantum communication n

Scientists at Peking University in China have cracked how to build long-distance quantum communication networks. In their recent research, the scientists demonstrated secure quantum communication over distances of 2,300 miles (3,700 km). 

Even as scientists and companies work on building error-proof quantum computers, the devices could end up being quite worthless if they cannot communicate with each other and transfer information over a quantum network. 

This is why researchers have also been working on developing secure quantum networks. Much of the effort has been on ensuring that the quantum networks can utilise existing fiber-optic networks. However, the development of such networks has been marred by prohibitive equipment costs and limited range. 

Shortcomings of QKD

Quantum Key Distribution (QKD) is regarded as the gold standard of quantum communication. The approach is known to be hack-proof, and even attempts to tap into the network are easily detected. These advantages of QKD also contribute to its shortcomings, as the network cannot serve multiple users simultaneously. 

Central to the QKD are the series of ‘trusted relay nodes’ which handle the quantum keys along the route. These nodes also introduce potential vulnerabilities into the quantum network. Researchers at Peking University decided to eliminate the relay nodes. 

QKD systems are also custom-built, making them prohibitively expensive. So, the researchers decided to base their approach on tech that can be easily manufactured at an industrial scale. 

How did they crack it? 

On the server side, the researchers introduced a super optical comb that generates ultra-stable laser lines at the same frequency. A low-frequency chip smaller than a fingernail, this comb enables devices to operate from an identical, non-wavering time base with a width as little as 40 hertz.  

On the client side, the team used 20 independent quantum transmitter chips with a complete suite of functions that enabled them to operate as telegraph operators at the quantum level.

The transmitters operated in pairs and received signals from the central comb. They then encoded the information into light pulses that could be sent over a fiber optic cable. 

The cable used in this setup was ~230 miles (370 km) long. In their experiments, the researchers found that the chip modulators achieved a 97.5 percent success rate. Since each pair could communicate over 230 miles, the aggregate networking capability across 20 chips was 2,300 miles (3,700 km). 

Both server and client-side chips demonstrated high performance and operational yield. Equally important, they were manufactured on industry wafers, making them highly scalable. 

The researchers are hopeful that their approach will help unlock intercity quantum networks that could support hundreds of users, without requiring any relays. More importantly, the use of standard fiber-optic cable in the setup shows that quantum networks will not require special cabling, making them viable in the long run. 

The team pointed out that their technology is not ready for deployment anytime soon and only works under carefully controlled laboratory conditions. To overcome this, the researchers have already begun the next phase of their work by incorporating single-photon detectors and optical frequency shifters onto the server chip and by expanding the number of microcomb channels, enabling them to serve more customers simultaneously.

🔗 Sumber: interestingengineering.com

📌 TOPINDIATOURS Eksklusif ai: MIT Researchers Unveil “SEAL”: A New Step Towards Se

The concept of AI self-improvement has been a hot topic in recent research circles, with a flurry of papers emerging and prominent figures like OpenAI CEO Sam Altman weighing in on the future of self-evolving intelligent systems. Now, a new paper from MIT, titled “Self-Adapting Language Models,” introduces SEAL (Self-Adapting LLMs), a novel framework that allows large language models (LLMs) to update their own weights. This development is seen as another significant step towards the realization of truly self-evolving AI.

The research paper, published yesterday, has already ignited considerable discussion, including on Hacker News. SEAL proposes a method where an LLM can generate its own training data through “self-editing” and subsequently update its weights based on new inputs. Crucially, this self-editing process is learned via reinforcement learning, with the reward mechanism tied to the updated model’s downstream performance.

The timing of this paper is particularly notable given the recent surge in interest surrounding AI self-evolution. Earlier this month, several other research efforts garnered attention, including Sakana AI and the University of British Columbia’s “Darwin-Gödel Machine (DGM),” CMU’s “Self-Rewarding Training (SRT),” Shanghai Jiao Tong University’s “MM-UPT” framework for continuous self-improvement in multimodal large models, and the “UI-Genie” self-improvement framework from The Chinese University of Hong Kong in collaboration with vivo.

Adding to the buzz, OpenAI CEO Sam Altman recently shared his vision of a future with self-improving AI and robots in his blog post, “The Gentle Singularity.” He posited that while the initial millions of humanoid robots would need traditional manufacturing, they would then be able to “operate the entire supply chain to build more robots, which can in turn build more chip fabrication facilities, data centers, and so on.” This was quickly followed by a tweet from @VraserX, claiming an OpenAI insider revealed the company was already running recursively self-improving AI internally, a claim that sparked widespread debate about its veracity.

Regardless of the specifics of internal OpenAI developments, the MIT paper on SEAL provides concrete evidence of AI’s progression towards self-evolution.

Understanding SEAL: Self-Adapting Language Models

The core idea behind SEAL is to enable language models to improve themselves when encountering new data by generating their own synthetic data and optimizing their parameters through self-editing. The model’s training objective is to directly generate these self-edits (SEs) using data provided within the model’s context.

The generation of these self-edits is learned through reinforcement learning. The model is rewarded when the generated self-edits, once applied, lead to improved performance on the target task. Therefore, SEAL can be conceptualized as an algorithm with two nested loops: an outer reinforcement learning (RL) loop that optimizes the generation of self-edits, and an inner update loop that uses the generated self-edits to update the model via gradient descent.

This method can be viewed as an instance of meta-learning, where the focus is on how to generate effective self-edits in a meta-learning fashion.

A General Framework

SEAL operates on a single task instance (C,τ), where C is context information relevant to the task, and τ defines the downstream evaluation for assessing the model’s adaptation. For example, in a knowledge integration task, C might be a passage to be integrated into the model’s internal knowledge, and τ a set of questions about that passage.

Given C, the model generates a self-edit SE, which then updates its parameters through supervised fine-tuning: θ′←SFT(θ,SE). Reinforcement learning is used to optimize this self-edit generation: the model performs an action (generates SE), receives a reward r based on LMθ′’s performance on τ, and updates its policy to maximize the expected reward.

The researchers found that traditional online policy methods like GRPO and PPO led to unstable training. They ultimately opted for ReST^EM, a simpler, filtering-based behavioral cloning approach from a DeepMind paper. This method can be viewed as an Expectation-Maximization (EM) process, where the E-step samples candidate outputs from the current model policy, and the M-step reinforces only those samples that yield a positive reward through supervised fine-tuning.

The paper also notes that while the current implementation uses a single model to generate and learn from self-edits, these roles could be separated in a “teacher-student” setup.

Instantiating SEAL in Specific Domains

The MIT team instantiated SEAL in two specific domains: knowledge integration and few-shot learning.

• Knowledge Integration: The goal here is to effectively integrate information from articles into the model’s weights.
• Few-Shot Learning: This involves the model adapting to new tasks with very few examples.

Experimental Results

The experimental results for both few-shot learning and knowledge integration demonstrate the effectiveness of the SEAL framework.

In few-shot learning, using a Llama-3.2-1B-Instruct model, SEAL significantly improved adaptation success rates, achieving 72.5% compared to 20% for models using basic self-edits without RL training, and 0% without adaptation. While still below “Oracle TTT” (an idealized baseline), this indicates substantial progress.

For knowledge integration, using a larger Qwen2.5-7B model to integrate new facts from SQuAD articles, SEAL consistently outperformed baseline methods. Training with synthetically generated data from the base Qwen-2.5-7B model already showed notable improvements, and subsequent reinforcement learning further boosted performance. The accuracy also showed rapid improvement over external RL iterations, often surpassing setups using GPT-4.1 generated data within just two iterations.

Qualitative examples from the paper illustrate how reinforcement learning leads to the generation of more detailed self-edits, resulting in improved performance.

While promising, the researchers also acknowledge some limitations of the SEAL framework, including aspects related to catastrophic forgetting, computational overhead, and context-dependent evaluation. These are discussed in detail in the original paper.

Original Paper: https://arxiv.org/pdf/2506.10943

Project Site: https://jyopari.github.io/posts/seal

Github Repo: https://github.com/Continual-Intelligence/SEAL

The post MIT Researchers Unveil “SEAL”: A New Step Towards Self-Improving AI first appeared on Synced.

🔗 Sumber: syncedreview.com