Markov Is Not “Memoryless”
It’s the First-Principles Trick That Makes Time, Decisions, and AI Actually Computable Continue reading on Towards AI »
Like
0
Liked
Liked
digitado
About digitado
Posts by :
![[P] VeridisQuo - open-source deepfake detector that combines spatial + frequency analysis and shows you where the face was manipulated](https://www.digitado.com.br/wp-content/uploads/2026/03/j51nr1pqomng1-YPF3wr.gif "[P] VeridisQuo - open-source deepfake detector that combines spatial + frequency analysis and shows you where the face was manipulated")
[P] VeridisQuo – open-source deepfake detector that combines spatial + frequency analysis and shows you where the face was manipulated
Hey everyone, My teammate and I just finished our deepfake detection project for university and wanted to share it. The idea started pretty simple: most detectors only look at pixel-level features, but deepfake generators leave traces in the frequency domain too (compression artifacts, spectral inconsistencies…). So we thought, why not use both? How it works We have two streams running in parallel on each face crop: An EfficientNet-B4 that handles the spatial/visual side (pretrained on ImageNet, 1792-dim output) […]
Like
0
Liked
Liked
Auditing AI Systems: A Practical Guide to Testing Models for Bias, Compliance, Security, and…
Auditing AI Systems: A Practical Guide to Testing Models for Bias, Compliance, Security, and Explainability Photo by Stephen Dawson on Unsplash Why accuracy alone is not enough and how organizations can audit AI systems before regulators, attackers, or users expose the failures. Artificial intelligence systems are now embedded in decisions that affect people’s lives credit approvals, fraud detection, hiring, underwriting, and customer support automation. But while AI adoption has accelerated rapidly, governance frameworks have struggled to keep up. Most […]
Like
0
Liked
Liked
![[D] Image Augmentation in Practice: In-Distribution vs OOD Augmentations, TTA, and the Manifold View](https://www.digitado.com.br/wp-content/uploads/2026/03/wsjk3r0xcmng1-dC3geb-280x140.jpg "[D] Image Augmentation in Practice: In-Distribution vs OOD Augmentations, TTA, and the Manifold View")
[D] Image Augmentation in Practice: In-Distribution vs OOD Augmentations, TTA, and the Manifold View
I wrote a long practical guide on image augmentation based on ~10 years of training computer vision models and ~7 years working on Albumentations. In practice I’ve found that augmentation operates in two different regimes: In-distribution augmentation Simulate realistic variation that could occur during data collection (pose, lighting, blur, noise). Out-of-distribution augmentation Transforms that are intentionally unrealistic but act as regularization (extreme color jitter, grayscale, cutout, etc). The article also discusses: • why unrealistic augmentations can still improve […]
Like
0
Liked
Liked
A unicorn-like Spinosaurus found in the Sahara
The Spinosaurus is a sail-backed, crocodile-snouted dinosaur that Hollywood depicted as a giant terrestrial predator capable of taking down a T. rex in Jurassic Park 3. Then they changed their mind and made it a fully aquatic diver in Jurassic World Rebirth—a rendering that was more in line with the latest paleontological knowledge. But now, deep in the Sahara Desert, a team of researchers led by Paul C. Sereno, a paleontologist at the University of Chicago, discovered new […]
Like
0
Liked
Liked
[P] NanoJudge: Instead of prompting a big LLM once, it prompts a tiny LLM thousands of times.
If you ask a traditional LLM to “rank these 1000 items,” it will hallucinate, lose the middle of the context, or just spit out cliches. I built an open-source tool called NanoJudge to fix this. It’s a pure-computation Rust engine that takes any list of items, hooks into any OpenAI-compatible local API (like vLLM or Ollama), and runs exhaustive pairwise tournaments (“Which is better: A or B?”). It then uses Bradley-Terry scoring and Bayesian MCMC sampling to compile […]
Like
0
Liked
Liked
The Death of the Consensus Computer Science Syllabus and the Rise of Innovation-First Learning
Why Traditional CS Learning is Breaking The clash between traditional education and rapidly advancing technology is no longer just a gap; it is now a complete break. For decades, schools have relied on slow cycles: years of theory, then controlled practice, all based on a syllabus that takes years to approve. In a world where AI brings new breakthroughs every week, waiting for consensus means falling behind. Traditional curricula are effectively “latency” in a high-speed world. By the time a […]
Like
0
Liked
Liked
From Iran to Ukraine, everyone’s trying to hack security cameras
For decades, satellites, drones, and human spotters have all been part of war’s surveillance and reconnaissance tool kit. In an age of cheap, insecure, Internet-connected consumer devices, however, militaries have gained another powerful set of eyes on the ground: every hackable security camera installed outside a home or on a city street, pointed at potential bombing targets. On Wednesday, Tel Aviv–based security firm Check Point released new research describing hundreds of hacking attempts that targeted consumer-grade security cameras […]
Like
0
Liked
Liked
If You’re Afraid of AI Replacing Developers, You’re Missing the Bigger Economic Reality
The Fear I like AI and have been working with the technology for over a decade. It’s been exciting, interesting, and inspiring — until recently. As I write this in mid-February of 2026, software stocks are in freefall and social media is inundated with videos of AI company CEOs predicting the looming collapse of the software industry. They tell us the story we’ve heard on loop for two years: “Most white-collar work will be fully automated within a […]
Like
0
Liked
Liked
Programmable Money as Policy: Designing Stablecoins for Emerging Markets
How smart contracts can encode fiscal discipline, aid conditionality, and trust Financial aid and monetary interventions in emerging markets are constrained less by intent than by execution. Capital is typically distributed through multilayered institutional pipelines involving governments, non-governmental organisations, correspondent banks, and local intermediaries. Each layer introduces delay, opacity, and opportunity for misallocation. By the time funds reach their intended recipients, their effectiveness is often diminished. These systems exhibit familiar failure modes. Disbursement delays undermine urgency. Leakage through […]
Like
0
Liked
Liked