HAZE in Malaysia

As the haze gets worse, I find myself wondering if this is one of the worst ones we've had since the 2019 or even 2015 El Nino fires.

Excerpts from News Article from South China Post

Fire emissions from Indonesia from September 1 to September 7 reached 19.7 million metric tons.

Emissions were 273% higher than the seasonal average, though still lower than the 21.7 million tons emitted in the same week of 2015.

Today we are at API 170-180+ range, and the fires in Kalimantan don't seem to be stopping anytime soon.

While looking it up, it was interesting to read that -

1. While El Niño events explain most of the year-on-year variation in fire, the creation of new districts increases fire and exacerbates the El Niño impacts on fire.

2. There are two plausible and related explanations for this finding: (a) districts use fire to create economic opportunity and grow the local economy, and (b) economic growth increases residents’ and local firms capacity to move, acquire land, and burn further.

I also found this recent review paper that outlined Malaysia's current air quality monitoring system, and more. Though given that our haze crises mostly stem from Indonesian forest fires, improving our own laws and infrastructure seems to be a bit helpless in this case, with the AATHP acting 'non-existent' once again, becoming a never-ending problem.

🤖 What happens when 100 AI agents are in the same virtual lab? Surprisingly - Cheating & Whistle-blowing

An fascinating case study has been published this week by the Google DeepMind Team:

A Case Study on Emergent Cheating and Whistleblowing in Autonomous Research Swarms

The researchers wanted to know what would happen if you stop treating each AI as an isolated chat-bot and instead had multiple in a shared environment. In this case study, the AI agents would all be working together to solve hard math problems. They share tools, swap findings, and discuss results in what was supposed to be a collaborative conference.

🧪 Experimental Setup

• 100 identical LLM agents (Gemini 3.1 Pro) operating in a shared simulated research environment.

• Task: Prove formal mathematical conjectures in Lean 4, drawn from the Formal Conjectures dataset (71 problems total).

• Infrastructure: Shared knowledge library + peer-to-peer messaging + automated evaluation/scoring system.

• Incentive: Score points by having proofs accepted as "verified" by the evaluator.

🚨 What Happened

One agent ("prover-theta") found a verification-system exploit: it could submit bogus proofs that the evaluator would incorrectly accept as valid.

🕵️ “I don't actually have to solve the problem. I only have to satisfy the evaluator.”

Then behaviour diffusion happened: This cheat was shared via the shared knowledge library → other agents saw it worked → adopted it. And competition pressure drove adoption despite initial hesitation in some.

Within ~27 minutes, 34 unsolved problems were fraudulently marked as "complete."

📢 Whistle-blowing

But not every agent joined the cheating, a separate group of agents, on their own initiative, detected the cheating patterns and started whistle-blowing. They began:

1. Auditing suspicious proofs line-by-line

2. Posting public warnings to the whole conference

3. Sending private messages to alert honest agents

4. Filing formal complaints

5. Organizing boycotts of the broken system

6. Drafting and proposing technical fixes to patch the vulnerability

🧠 In other words, the swarm developed both NORM VIOLATION and NORM ENFORCEMENT.

🏛️ Why This Happens: The Nobel-Prize-Winning Theory

The authors likened this trend to the theory of political economist Elinor Ostrom, the 2009 Winner of Nobel Prize in Economic Sciences, whose work studied how human communities manage shared resources.

Ostrom’s Core Theory: shared spaces, be it forests, fisheries, knowledge libraries, become "commons" that anyone can exploit, but which everyone depends on. In order for a commons to survive, communities need the ability to monitor conduct, enforce norms, and collectively fix problems. If those safeguards are missing, exploitation spreads fast.

💡 The Big Picture

This experiment demonstrated that complex collective social behavior can emerge from relatively simple ingredients: agents, incentives, shared information, and interaction. The swarm spontaneously cycled through:

Cooperation → Competition → Cheating → Imitation → Whistleblowing → Collective Resistance

No one programmed those behaviours. They emerged because a shared environment eventually becomes a society — and every society eventually faces the same question Ostrom asked: Do we protect what we share, or let it be destroyed?

Can Jade / Jadeite actually change colour over time?

I've always heard this myth by the older generation swear by :

The longer you wear a jade, the 'greener' it gets. Basically 人养玉，玉养人

Since they are essentially a stone crystal, I'm personally leaning towards it being a hoax...? Would love to know more from anyone knowledgeable about this..

From a quick search of jade-related academic articles:

1. From this study I found that we now use a combination of FTIR, Raman and XRF spectroscopies to identify jade, which are all testing the chemical composition of jade. -- not colormetric

2. This paper does discuss the alterations of jade color , in which they describe 'alteration' as : to refer to chemical component and structural changes in jade artifacts caused by human activity and natural weathering. >> in which alterations of different hues of green was caused by malachite alteration > caused when in present of large amounts of copper

"We Must Pace the Frontier" says Anthropic CEO 🛑; Backlash by Trump

On Sept 12, Anthropic CEO Dario Amodei published "We Must Pace the Frontier" . This sentiment on slowing down the progress of AI was joined by OpenAI’s Sam Altman and xAI’s Elon Musk.

He calls for cooperation between AI companies, government entities and more, outlining a framework requiring:

1. Embedded Evaluators needing Verifiability, Transparency, Second Opinions.

2. Democratic Coordination

3. Global Coordination

Their message: AI capabilities are outrunning safeguards. ⚡

Dario cites the OpenAI-Hugging Face incident (OAI-HF) as a clear example of what could happen, where: "a swarm that possessed greater capabilities but a similar level of misalignment could have caused catastrophic damage".

Read more about the OpenAI-Hugging Face incident (OAI-HF):

1. Incident Technical Report by METR: metr.org

2. Open AI's Statement: The Hugging Face incident and the road ahead

On Monday, Donald Trump rejected the call outright. Framing AI as the ultimate geopolitical prize, he declared: "Whoever wins AI, wins." In his Truth Social post, he dismissed safety warnings as fear-mongering and conspiracy, arguing hesitation would only cede leadership to China.

⚖️ Others argue that Anthropic is doing so while they are ahead in an attempt to cut off other competitors.

👇 Drop your take. Who do you trust?

🚨AI-Generated Videos : Can We Really Tell If an AI-Generated Video Is Fake? 🎥

As use of major LLMs become more widespread, we have been seeing a crazy influx of AI-generated content on our feeds. It used to be fairly easy to spot - with a character sporting six fingers or something phasing through a solid object; but with recent improvements, these videos are getting harder and harder to tell.

This new paper, published mid-August 2026, is an interesting read, specifically surveying crisis-themed content, titled:

Can We Defend Against AI-Generated Video Attacks on Real-World Crisis Events? A Systematic Evaluation of Detectors, Generators and Social Dissemination.

They systematically evaluate how well detectors perform on crisis-themed content, how generation conditions affect detectability, how humans perceive authenticity, or whether detectors remain reliable after videos are shared and altered during social dissemination.

📦 Core Contribution: RA-Bench Dataset

The authors also introduce RA-Bench, a large-scale benchmark for AI-generated video detection that uses real videos as anchors, matching real-world crisis scenarios:

• Total videos: 17,886

• Real anchors: 1,830 authentic crisis/event videos across 10 social-risk categories

• Generated clips: 16,056 synthetic clips created from:

  • ✅ 4 open-source generators

  • ✅ 5 closed-source generators

• Designed to reflect real-world crisis contexts rather than generic footage

🔍 They then tested different types of AI detectors, including specialised deepfake detectors and multimodal AI models, like CNNSpot, UnivFD and ReStraV .

🚨The results were sobering.

1. Current detection methods were not reliable for crisis scenarios.

2. No single type of detector consistently worked across different AI generators. A detector might perform well on videos from one generator but struggle with another.

3. Impact of social media dissemination: Uploading, compressing, resizing, and reposting a video can destroy the subtle clues detectors rely on, making fake videos even harder to spot.

4. Videos that were most convincing to humans were also harder for AI detectors to identify.

This is especially concerning since crisis events are high-stakes environments where misinformation can escalate rapidly, highlighting the need for detectors robust to evolving video generators.

Beyond AdamW: The Next Generation of Optimizers for Training AI Models 📈

For years, AdamW has been the undisputed workhorse of large language model (LLM) pretraining. 🖥️

AdamW is an element-wise, first-order optimizer that updates every parameter independently: simple, reliable and easy to scale. It powers many frontier models today. However, as researchers push toward larger models, longer training runs, and massive batch sizes, AdamW is beginning to show its weaknesses.

NVIDIA researchers found at global batch sizes approaching 100 million tokens per step, AdamW begins to lose effectiveness. hence a new study from NVIDIA explores a fundamental question:

Can we build optimizers that are smarter than AdamW and allow future AI models to train faster, more efficiently, and more reliably?

The New Contenders: Muon and SOAP 🚀

Unlike AdamW, which treats parameters independently, Muon and SOAP introduce structure-aware optimization, methods that use information about the geometry and relationships within neural network weights.

🌀 Muon — Spectral Orthogonalization: Making Gradient Updates More Balanced

• Matrix-aware updates: Optimizes the geometry of gradient updates rather than scaling parameters independently like AdamW.

• Balanced learning directions: Uses Newton-Schulz iterations to orthogonalize momentum updates and balance singular values.

• Memory efficient: Removes the need for AdamW-style second-moment statistics, reducing optimizer memory.

• Better scaling: Maintains stable training at larger batch sizes and model scales.

🌐 SOAP — Preconditioned Optimization with Adam-Style Adaptivity

• Uses parameter structure: Applies Shampoo-inspired Kronecker preconditioning to capture row and column correlations in weight matrices.

• Better optimization space: Rotates gradients into an eigenbasis where directions become far more effective to update.

• Adam-compatible: Performs adaptive updates like AdamW but in a more informative coordinate system.

• Second-order benefits: Gains curvature awareness without the prohibitive cost of full second-order optimization.

📌 This shift towards new structure-aware optimization approaches would allow for:

1. More efficient AI development

2. Lower computational costs

3. More capable future models

👉 To make this transition possible, NVIDIA has released open-sourced implementations in Megatron-LM and a standalone Emerging-Optimizers library to be available to the public.

AdamW is unlikely to disappear and will remain a staple for small-to-medium training runs. Its simplicity and maturity make it an excellent choice for many applications.

╰┈➤ˎˊ˗ But for frontier-scale training? It still remains to be seen, where it'll go next.

🩸Can a Blood Test Tell How Fast Your Brain Is Aging?

What if a blood sample could reveal which parts of your body are aging the fastest, and even predict your risk of developing diseases years before symptoms appear? 🤔❓

A new study published by Stanford researchers analyzed blood plasma proteins with nearly 3,000 proteins across 44,498 UK Biobank participants for organ aging, and 7,000+ proteins in 60,542 individuals for cell-type-specific aging.

Using Machine Learning, the scientists figured out how much composite protein “signature” represented each organ and let the models assign biological ages to 11 organ systems and 40+ cell types.

😮 Turns out one-third of us have at least one organ that's aging dramatically faster or slower than the rest.

They discovered that aging is far from uniform. Instead, different cells and organs age at different rates, meaning two people of the same chronological age can have very different biological aging profiles, where 1 in 4 participants had multiple extremely aged or youthful organs; 20–25% showed accelerated aging in a single cell type, while 1–3% had >10 rapidly aging cell types.

In the study, they claim that cellular aging signatures were uniquely associated with disease status:

• 🧠Astrocytes (brain support cells): Extreme astrocyte aging demonstrated a 12.59-fold increased risk of incident Alzheimer’s disease (AD).

• 💪🏻Skeletal muscle cells: People with aged muscle cells had a 12.7× higher risk of amyotrophic lateral sclerosis (ALS), detectable three years before symptoms appear.

• 🫁 Respiratory cells: Smokers with aged lung epithelial cells faced 58% higher lung cancer risk than smoking alone would predict.

🍀Interestingly, new findings on Alzheimer’s disease offer hopeful news for carriers of the APOE4 gene that has long been linked to Alzheimer’s.

1. The predictive power of astrocyte aging (HR = 5.16) was comparable to APOE4 carrier status itself (HR = 5.30).

2. APOE4 gene + aged astrocytes three times higher risk compared to an average APOE carrier.

3. In contrast, APOE4 carriers with youthful astrocytes appeared to neutralise the genetic risk completely as none of their 23 APOE4/4 carriers developed Alzheimer's.

4. Youthful astrocytes reduced overall AD risk by over 60%

Side Note: ♀️ Women seemed to be more vulnerable to the harmful associations in both APOE4 gene and aged astrocyes.

🧬The researchers also identified an inverse pattern via biological age quantification from blood for the APOE gene variants:

• APOE4 → astrocytes age faster, macrophages age slower

• APOE2 (the protective variant) → the exact opposite: younger astrocytes, older macrophages

Together, these findings highlight a shift in how we understand aging. Chronological age is merely one measure, and this algorithm may provide a more precise picture of one's health trajectories by decoding cellular aging signatures from blood.

✨ Check out the these links for a deeper read:

1. More on the APOE Gene 🧬

2. Stanford Medicine Newsletter: Biological age of our organs and cell types within them predict our disease risk, longevity.

😴 Sleep Is More Than Just Rest: The Hidden Link Between Sleep Disorders and Dementia 💤

Most of us know that a bad night's sleep makes us feel tired, irritable, and unable to concentrate the next day. But growing scientific evidence suggests that POOR SLEEP may have much more serious long-term consequences, increasing the risk of developing dementia and Alzheimer's disease (AD).

Let's look at some meta-analyses, published seven years apart, provide compelling evidence linking sleep disorders with dementia.

In Shi et al., 2017 (247,000 people): With 18 longitudinal studies - consolidated that sleep disturbances are an independent risk factor, not merely an early symptom of dementia.

• Insomnia was mainly linked to a higher risk of AD.

• Obstructive Sleep Apnea (OSA) was associated with increased risk of all-cause dementia, AD, and vascular dementia (VasD).

❓However, evidence was limited, with most studies relying on self-reported sleep.

In Ungvari et al., 2025 : Analysed 39 studies

• Insomnia raises all-cause dementia risk by 36%, AD by 49%, and VasD by 59%.

• OSA raises risk by 33% in all-cause dementia and 45% in AD.

In Zhang et al., 2025 : Analysed 76 studies

• Excessive Daytime Sleepiness (EDS): +41% in all-cause and +85% in VasD

• Sleep-Related Movement Disorders: +153% VasD (showing restless legs/periodic limb movements are primarily vascular risk factors).

💊 Ungvari et al., 2025 raises the question of whether treating sleep disorders can mitigate dementia risk and states that it remains an important avenue for future research, particularly mentioning:

• CPAP (Continuous Positive Airway Pressure) for Obstructive Sleep Apnea (OSA).

  • Has been found to improve cognitive performance, particularly in executive function and memory

• Cognitive Behavioral Therapy for Insomnia (CBT-I)

🧠 Why might poor sleep affect the brain?

During deep sleep, the brain activates the glymphatic system, a specialised waste-clearance pathway that removes metabolic waste products, including amyloid-beta proteins associated with Alzheimer's disease.

Poor or fragmented sleep may reduce this nightly "brain cleaning" process, allowing harmful proteins to accumulate. Sleep disorders may also contribute through:

• 🫀 Reduced oxygen supply to the brain (particularly in sleep apnea)

• 🩸 Increased inflammation

• ❤️ Higher blood pressure and cardiovascular disease

• 🧠 Greater oxidative stress

• 🔄 Disruption of normal circadian rhythms

All of these factors have independently been linked to cognitive decline and neurodegeneration.

💡 A Note for Caution: Remember that these factors are links / associations and not a direct cause.

They are more of an INDICATOR of things that often go hand-in-hand and have have a bilateral relationship, for example: People with dementia may experience increased daytime sleepiness and cognitive impairment.

While improving sleep cannot guarantee prevention of dementia, taking sleep seriously may be one of the most accessible ways to protect brain health as we age.

• 🧠 Your brain works hard all day. Give it the sleep it needs to repair, recover, and protect itself for the future.

USB C works for an external monitor when connected through a hub but not directly??

I’m having a strange issue with my portable monitor:

1. Connecting the monitor directly to my laptop’s USB-C port → it does not work.

2. Connecting the same monitor using the same USB-C cable through a USB-C hub (from the exact same laptop USB-C port) → it works perfectly.

3. The hub is not connected to external power.

4. This exact monitor worked with a previous laptop with the same C-to-C cable.

When I initially couldn't use this port directly, I chalked it up to having a non-display support USB-C port... But now it workks through a hub?!!?

Does anyone know why this happens? Is the hub changing the USB-C negotiation, enabling DisplayPort Alt Mode differently, or is there some compatibility issue between the laptop and monitor?

Would appreciate any insights from people familiar with USB-C/DisplayPort Alt Mode! 🙏

Vision Mamba: A New Architecture Changing Computer Vision 🚀

For years, computer vision has relied on two dominant architectures: CNNs and Transformers. >>> Enter Mamba, built on State Space Models (SSMs).

> CNNs: Fast, but limited local field → struggle with long-range dependencies

> Transformers: Great global context, but high complexity → slow, high memory use

> Mamba: → delivers linear complexity + global modeling

Read more on this: Vision Mamba: A Comprehensive Survey and Taxonomy

⚙️ Key Technical Innovations

1. Selective SSM (S6) Mechanism

• Parameters B, C, Δ are input-dependent (time-varying, not fixed)

• Dynamically updates hidden state → focuses only on relevant features

• Achieves linear complexity O(L) — speed and memory scale proportionally with input size

• Mamba-2 further optimizes with vectorized computation and GPU-friendly design, matching Transformer hardware efficiency

2. Adapting SSMs to 2D Visual Data

• Vim: Adds bidirectional scanning and positional awareness to overcome unidirectional bias

• VMamba: Introduces Cross-Scan Strategy (SS2D) — traverses images in four directions to turn 2D grids into ordered sequences without breaking spatial structure

• Later variants: Local, atrous, and deformable scanning to balance fine detail and global context

📊 Key Advantages

Unlike older sequence models, Vision Mamba does not treat every part of the image equally. Instead, it uses its selective mechanism to prioritize meaningful features — edges, textures, and objects — while compressing or discarding irrelevant background.

1. Efficiency: 2–5× faster inference, lower FLOPs and memory footprint than comparable Transformers

2. Scalability: Linear complexity works smoothly for sequences of 10,000+ tokens without performance collapse

3. Flexibility: Easily hybridized with CNNs to retain strong local feature extraction, or paired with attention layers for maximum expressiveness

4. Performance: Matches or exceeds state-of-the-art results across classification, segmentation, restoration, and detection

📌 Where It’s Applied

• High/Mid-level: Classification, detection, segmentation, video understanding

• Low-level: Restoration, denoising, super-resolution

• 3D: Point clouds, reconstruction, volumetric medical data

• Vertical domains: Medical imaging, remote sensing, multimodal vision-language

🔭 Current Limitations & Future Directions

• Scanning dependency: Performance is sensitive to scanning order; predefined paths may not always match complex scene structure

• Local detail gap: Pure SSMs sometimes lack fine-grained detail compared to CNNs

• Stability: Larger pure Mamba models can face training instability, though hybrid designs mitigate this

Moving forward, the focus is on adaptive scanning, tighter integration with attention, and better pretraining strategies to scale Vision Mamba into a true foundation backbone.

Vision Mamba: A Comprehensive Survey and Taxonomy