Neural Nets for Combinatorial Optimization
Pointer Neural Nets and Graph Neural Nets with Beam Search to construct approximate TSP tours
Read More$P(A|B) = \frac{P(B|A)P(A)}{P(B)}$
$\vec{x} = [x_1, x_2, ..., x_n]^T$
$\nabla f(x) = [\frac{\partial f}{\partial x_1}, ..., \frac{\partial f}{\partial x_n}]$
$y = \sigma(w \cdot x + b)$
$H(X) = -\sum_{i} P(x_i) \log P(x_i)$
$f(x) = \frac{1}{1+e^{-x}}$
$\mathbf{X} \sim \mathcal{N}(\mu, \Sigma)$
$\hat{y} = \beta_0 + \beta_1 x_1 + \ldots + \beta_p x_p$
$E[X] = \int_{-\infty}^{\infty} x f(x) dx$
$\text{MSE} = \frac{1}{n} \sum_{i=1}^{n} (y_i - \hat{y}_i)^2$
$p(\theta | \mathcal{D}) \propto p(\mathcal{D} | \theta) p(\theta)$
$\min_\theta \mathcal{L}(\theta) = -\sum_{i=1}^{N} y_i \log(\hat{y}_i)$
$\text{KL}(P || Q) = \sum_x P(x) \log \frac{P(x)}{Q(x)}$
$\text{Var}(X) = E[(X - \mu)^2]$
$\nabla_\theta J(\theta) = E_{\pi_\theta}[\nabla_\theta \log \pi_\theta(a|s) \cdot Q(s,a)]$
$\phi(x) = \frac{1}{\sigma\sqrt{2\pi}} e^{-\frac{1}{2}\left(\frac{x-\mu}{\sigma}\right)^2}$
$\mathbf{W}^* = \arg\min_{\mathbf{W}} ||\mathbf{Y} - \mathbf{X}\mathbf{W}||_F^2$
$A_t = r_t + \gamma V(s_{t+1}) - V(s_t)$
$\mathcal{F} = -\text{tr}(\mathbf{p} \log \mathbf{p})$
$\text{ELBO} = E_{q}[\log p(x,z)] - E_{q}[\log q(z)]$
$\mathbf{H} = \nabla^2 f(\mathbf{x})$
$S = k \log W$
$\text{Cov}(X,Y) = E[(X-E[X])(Y-E[Y])]$
$\hat{\beta} = (X^TX)^{-1}X^Ty$
∇L(θ) → 0: Tomáš Kozák
Senior Data Scientist & AI Engineer
Background
FIS VŠE
Master's in Data & Analytics for Business
Thesis: MLOPTIFLOW - Dynamic MLOps Framework with Integrated CLI for Automated ML Project Inception, Kafka-Driven Real-Time Model Monitoring, and Adaptive Canary Deployment Architectures
You can listen to the thesis podcast generated by NotebookLM:
MIT IDSS
MicroMaster's in Statistics & Data Science
MIT Institute for Data, Systems, and Society
Stealth
Founding AI Engineer
Currently in stealth mode
knowdroids.ai
Senior AI Engineer
Autonomous AI Agents development
Sympulse
Senior Data Scientist & AI Engineer
Custom AI / ML applications development
EY Parthenon
Senior AI Engineer
AI applications development for Risk Management
EY
Data Scientist
Custom AI / ML applications development
Revolt BI
Data Engineer
Custom data engineering & BI solutions development
Primary Stack
Python
PyTorch
PyTorch Lightning
Captum
TorchTune
FastAPI
LangGraph
Azure
Docker
Wanderlust
Exploring, one place at a time
Latest from the Blog
My random AI / ML writings about topics I want to explore but usually don't have time for
Neural Self-Organization: Growing Patterns with Neural Cellular Automata
Intro to differentiable morphogenesis and emergent self-repair
Read More