top of page
  • Writer's picturePeter K F Cheung SBS


  1. FADE IN

  2. Act 1


  4. A few SWIMMERS. PETER enters the pool.

  5. PETER (V.O.): Today is May 4th. It reminds me that this pool opens on May 1st. For the past three days, I swam elsewhere.

  6. PETER swims past a COACH and his young PUPIL.

  7. PETER (V.O.): No one has taught me how to swim.

  8. Swimming breast strokes to one end.

  9. PETER (V.O.) (Cont'd): When young, I played in the water. Then, I observed how others swam.

  10. Turning around, Peter swims to the other end.

  11. PETER (V.O.) (Cont'd): I was curious how others could do it so well.

  12. Swimming freestyle.

  13. PETER (V.O.) (Cont'd): I practise nearly daily, trying to do better.

  14. Act 2


  16. INT. STUDY - 11:00 (Earlier today)

  17. Peter is reading some papers on screen.

  18. PETER (V.O.): What's "federated learning" in AI?

  19. We see Peter posting the question to ChatGPT.

  20. PETER (V.O.) (Cont'd): It's a machine learning technique that enables multiple devices to collaboratively train a shared machine learning model without sharing their data with each other.

  21. Pausing.

  22. PETER (V.O.) (Cont'd): I used to think that in machine learning, a large dataset is collected in one location and the model is trained on the data.

  23. Reading.

  24. PETER (V.O.) (Cont'd): In federated learning, the data remains on the individual devices and the model is trained locally on the individual device.

  25. Reading.

  26. PETER (V.O.) (Cont'd): The updated models for each device are then sent to a central server, where they're aggregated to create a new, improved model.

  27. Thinking.

  28. PETER (V.O.) (Cont'd): To allow for privacy and secrecy, this federated learning is really smart!

  29. Pausing.

  30. PETER (V.O.) (Cont'd): I want to know more.

  31. Posing more questions to ChatGPT, Peter studies and analyses them.

  32. PETER (V.O.) (Cont'd): Ah, "Federated Averaging" (FedAvg) is even more interesting.

  33. Pausing.

  34. PETER (V.O.) (Cont'd): It trains a global model by aggregating the local models trained on different devices in a decentralised network.

  35. Thinking.

  36. PETER (V.O.) (Cont'd): Its application includes natural language processing. Hmm, I believe I'm enjoying this benefit.



  39. PETER (V.O.) (Cont'd): Hours ago, I generated an opportunity to talk about AI with knowledgeable experts.

  40. Smiling.

  41. PETER (V.O.) (Cont'd): While they know federated learning readily, they haven't heard the idea of FedAvg.

  42. Pausing.

  43. PETER (V.O.) (Cont'd): I don't think the FedAvg is cooked up by ChatGPT, as the concept sounds like Federated Learning 2.0 to me.

  44. Pausing.

  45. PETER (V.O.) (Cont'd): I admit that I've no background in any science and I only learn about federated learning and others today.

  46. Peter swims past a female SWIMMER.

  47. PETER (V.O.) (Cont'd): Today, I also read about the story of a famous Hollywood actress in the 1960s. She said the "her" in epic movie was gone.

  48. Pausing.

  49. PETER (V.O.) (Cont'd): In an album link I received today, I can recognize my former colleagues readily, but not myself!

  50. Act 3


  52. Peter swims a few butterfly strokes.

  53. PETER (V.O.) (Cont'd): I practise lifelong learning. I seize moments and convert them into opportunities for all to learn more and better.

  54. Taking a rest by the jacuzzi.

  55. PETER (V.O.) (Cont'd): The core of my strength is my capacity to make something value-added out of something.

  56. Swimming a few more butterfly strokes.

  57. PETER (V.O.) (Cont'd): Any turning points can come at the unexpected times and in unexpected ways. May 4th is always a good day!

  58. Peter leaves.

  59. FADE OUT

  60. THE END

12 views0 comments

Recent Posts

See All


bottom of page