Proximal Policy Optimization in Action: Real-Time Pricing with Trust-Region Learning
Author(s): Shenggang Li Originally published on Towards AI. Photo by Tesa Kimbal on Unsplash Introduction Every time a customer opens an app or website, the platform must set a surcharge in milliseconds to balance rider supply, demand spikes, and weather. Simple if-then …
Hybrid Model-Based RL for Intelligent Marketing: Dyna-Q Meets Transformer Models and Bayesian Survival Priors
Author(s): Shenggang Li Originally published on Towards AI. A theory-to-practice study on profit-driven customer re-engagement in e-commerce using BG/NBD-augmented Attention and budget-aware roll-outsPhoto by J Di on Unsplash We built a next-gen coupon engine fusing three techniques: a Bayesian survival model for …
Temporal Graph Neural Networks for Multi-Product Time Series Forecasting
Author(s): Shenggang Li Originally published on Towards AI. Modeling Cross-Series Dependencies and Temporal Dynamics in Retail Supply-Chain DataPhoto by Laszlo Biro on Unsplash This paper uses a supply-chain scenario — forecasting daily sales for multiple products — to explore Graph Neural Networks …
Plug-and-Play Reinforcement Learning for Real-Time Forecast Recalibration
Author(s): Shenggang Li Originally published on Towards AI. Updating legacy ARMA sales models with a PPO residual corrector — no full retrain requiredPhoto by Anders Jildén on Unsplash When I build a time-series model — say an ARMA trained on last season’s …
Distill-then-Detect: A Practical Framework for Error-Aware Machine Learning
Author(s): Shenggang Li Originally published on Towards AI. Leveraging Teacher Uncertainty, Student Distillation, and Conformal Calibration to Diagnose and Flag High-Risk PredictionsPhoto by Sigmund on Unsplash Even the most advanced neural networks or boosting algorithms sometimes stumble on a small but critical …
From Static to Dynamic: Evolving Bayesian Network Thinking for Real-World Applications
Author(s): Shenggang Li Originally published on Towards AI. Applied Bayesian Networks: Bridging Theory, Modeling, and Forecasting in PracticePhoto by Abi Ghouta Timur on Unsplash Imagine you’re a supply-chain manager trying to predict equipment failures before production halts. Begin by mapping key factors …
Hybrid Attention for Binary Sequence Forecasting
Author(s): Shenggang Li Originally published on Towards AI. Combining n-Gram Embeddings, Count-Aware Self-Attention, and Recency-Weighted ARMA for Multi-Horizon Distributional PredictionsPhoto by Pasqualino Capobianco on Unsplash I tackle pure binary time series forecasting by converting complex signals into 0/1 patterns — stock up/down, …
Designing Customized and Dynamic Prompts for Large Language Models
Author(s): Shenggang Li Originally published on Towards AI. A Practical Comparison of Context-Building, Templating, and Orchestration Techniques Across Modern LLM FrameworksPhoto by Free Nomad on Unsplash Imagine you’re at a coffee shop, and ask for a coffee. Simple, right? But if you …
A Novel and Practical Meta‑Booster for Supervised Learning
Author(s): Shenggang Li Originally published on Towards AI. A Stacking‑Enhanced Margin‑Space Framework for Dynamic, Loss‑Driven Ensemble Updates in Classification and RegressionPhoto by Thorium on Unsplash Ensemble methods thrive on diversity, yet most frameworks exploit it sequentially (boosting) or statically (stacking). We introduce …
Adaptive Decay-Weighted ARMA: A Novel Approach to Time Series Forecasting
Author(s): Shenggang Li Originally published on Towards AI. Integrating Recency-Based Loss Weighting and Seasonal Feature Tuning for Enhanced Predictive AccuracyPhoto by Haberdoedas II on Unsplash Time series forecasting is both fascinating and challenging. It’s fascinating because accurate predictions can directly inform better …
Beyond Simple Inversion: Building and Applying Inverse Neural Networks
Author(s): Shenggang Li Originally published on Towards AI. Theory, training tricks, and real‑world case studies — solving multi‑root equations and beyondPhoto by Marisa Harris on Unsplash Inverse problems ask a fundamental question: Given the output y, what was the input x? Traditional …
Reinforcement Learning-Enhanced Gradient Boosting Machines
Author(s): Shenggang Li Originally published on Towards AI. A Novel Approach to Integrating Reinforcement Learning within Gradient Boosting Internal Optimization for Superior Predictive PerformancePhoto by Austin Neill on Unsplash In this post, I demonstrate how reinforcement learning (RL) can directly enhance the …
Adaptive Multi-Teacher Distillation for Enhanced Supervised Learning
Author(s): Shenggang Li Originally published on Towards AI. A Novel Approach for Dynamically Combining Multiple Predictive Models into a Lightweight High-Performance Student ModelPhoto by Damon Hall on Unsplash In practical supervised learning, using a single predictive model like XGBoost, LightGBM, or Random …
Reimagining Diffusion Models: Autoregressive Priors for Efficient Initialization
Author(s): Shenggang Li Originally published on Towards AI. Exploring a Novel Approach to Diffusion Initialization with Intuitive Illustrations, Applications This member-only story is on us. Upgrade to access all of Medium. Photo by Gary Fultz on Unsplash Diffusion models have become a …
Practical Guide to Distilling Large Models into Small Models: A Novel Approach with Extended Distillation
Author(s): Shenggang Li Originally published on Towards AI. Comparing Traditional and Enhanced Step-by-Step Distillation: Adaptive Learning, Cosine Similarity, and Curriculum-Based Rationale Supervision This member-only story is on us. Upgrade to access all of Medium. Photo by Thorium on Unsplash In this paper, …
Popular posts
Updates
Recent Posts
