News for hackers. Stuff that matters.

haifeng

haifeng on Hacker News

Re:E-means: 10x faster k-means at equal loss on Tuesday November 10, 2026 @02:50PM (#50445393)
I add a Java implementation of E-Means in SMILE: https://github.com/haifengl/smile/blob/master/core/src/main/...

Let me know if you hit any issues.

Re:E-means: 10x faster k-means at equal loss on Tuesday November 10, 2026 @12:50PM (#50443664)
Look deeper, I am not sure that we can combine BBD-tree with E-Means. BBD-tree is a data structure built on the whole dataset, while the core idea in E-Means is mini-batch with exponential moving average updates.
Re:E-means: 10x faster k-means at equal loss on Tuesday November 10, 2026 @10:18AM (#50441611)
Lloyd is not the fastest algorithm for k-means for very long time. I am curious how e-means compares with more advanced algorithms such as BBD-tree based implementation. SMILE has a k-means implementation with both BBD-tree and llyod algorithm. BBD-tree approach is often 10x faster. Wonder if e-means is better than BBD-tree.

See https://www.aihalo.dev/api/java/smile/clustering/KMeans and https://www.aihalo.dev/api/java/smile/clustering/BBDTree

Maybe we combine the ideas of e-means and BBD-tree?

Show HN: AutoML Agents

Posted by haifeng on Tuesday May 05, 2026 @09:05AM
from the show-hn dept.

https://github.com/haifengl/smile/tree/master/studio

SMILE Studio is an IDE for machine learning and data science. It combines an interactive notebook and AI-powered agents in a single, modern desktop application. The AutoML Agent is capable of automating the end-to-end workflow of building, training, and deploying machine learning models. You may try it with the prebuilt package. You may also guide the agent to build ML/AI solutions step by step...

( Read More... | no comments | 3 points | on HN )

Show HN: Smile-Serve – Inference Server for ML, ONNX, and LLM

Posted by haifeng on Monday May 04, 2026 @06:02PM
from the show-hn dept.

https://github.com/haifengl/smile/tree/master/serve

SMILE Serve is a production-ready inference server built on [Quarkus](https://quarkus.io/) that brings together three complementary inference capabilities on the JVM: - **Classic ML**: `/api/v1/models` for serialized SMILE models (`.sml`) - **ONNX Runtime**: `/api/v1/onnx` for any model in the ONNX open format (`.onnx`) - **LLM Chat**: `/api/v1/chat` for Llama 3 chat completions A...

( Read More... | no comments | 2 points | on HN )

Older »