Industrial AI·10 min read·New
Meta Acquires Manus: The AI Agent Wars Have Begun
Analysis of Meta's acquisition of Manus AI, what it signals for the AI agent market, and how this reshapes the competitive landscape between Big Tech players.
About
Statistical ML background with focus on industrial applications. Interested in bringing research to practice.
Statistical Foundation
Probability theory, optimization, and statistical inference. Theoretical grounding for reliable ML systems.
AI & Deep Learning
Transformers, computer vision, time-series analysis. Published research and practical implementations.
Industrial Automation
Applying ML research to real-world systems. From theory to production.
Featured
Research publications, academic programs, and invited presentations.
Session
Enhancing AI with Advanced Sampling and Data Splitting Techniques for Big Data
Hilbert Split
Proposed a robust data splitting methodology for ML based on statistical ML theory.
Key Contribution
Theoretical framework for reliable train-test splits that preserve statistical properties.
Audience
International gathering of statisticians and data scientists.
Rate-Distortion Theory
Quantization-aware matrix multiplication through the lens of rate-distortion theory for efficient model compression.
Theoretical Lower Bounds
Deriving theoretical lower bounds for Kernel Ridge Regression (KRR) and Gradient Descent optimization.
High-Dimensional Statistics
Statistical analysis in high-dimensional regimes where classical theory breaks down.
Research Community
Intensive program with leading researchers in theoretical ML.
Problem
Detecting 'bradykinesia' - subtle movement hesitations in Parkinson's patients.
Solution
LSTM-FCN system detecting movement hesitations and amplitude decay in finger-tapping videos.
Result
80.3% accuracy on 1,396 videos from 310 patients.
Innovation
Hierarchical classification for temporal patterns of parkinsonian movement.
Research
Current and past research projects.
Ongoing
Sensitivity Analysis
Building scalable variable selection and ranking algorithms for noisy datasets.
Variable SelectionStatisticsML Practice
2025Ongoing
AI for Radiotherapy
AI-driven radiotherapy optimization using sensitivity analysis for treatment planning.
HealthcareSensitivity AnalysisAI
2025Ongoing
Hilbert Split
Robust data splitting methodology for ML based on statistical theory. Presented at JSM 2025.
StatisticsData SplittingML Theory
2025Blog
Theory of AI · AI Agent · Industrial AI
Theory of AI·4 posts
Quantization for Matrix Multiplication
Why is matrix multiplication the bottleneck in LLM inference? We formulate the rate-distortion problem for matmul and review existing weight-only quantization methods.
1. Problem Formulation...2. Why Classical Method...3. The Optimal Rate-Dis...+1 more
AI Agent·8 posts
Claude Cookbooks Study
Introduction to Anthropic's Claude Cookbooks repository. Covers classification, summarization, and text-to-SQL capabilities.
1. Overview & Basic Cap...2. RAG & Contextual Emb...3. Tool Use & Function ...+5 more
Get in Touch
Have a question or want to collaborate? Send a message.