October 12 2021 AI, Jupyter, ELIZA, Node-RED
Darren began by sharing what he had been looking at in machine learning and AI: Siri/Alexa/Picroft and a number of personal projects including
- MENACE (Matchbox Educable Noughts and Crosses Engine)
- draughts
- NIM (see this PDF)
- Jupyter notebooks, formerly IPython Notebooks, a web-based interactive environment for creating notebook documents; it is a bit like Zim in that you can import images, run programs, etc.
Darren demonstrated opening a Jupyter document, running a program, graphing the results and documenting it; you can also inspect the code.
Bernard commented that Jupyter encourages you to put everything into global variables rather than using local variables.
Darren noted that AI needs a complicated database with specific datatypes, encompassing both categorical, often qualitative, data and numerical data, though John commented that IBM had developed the Watson computer to search very large quantities of unstructured data and find answers very quickly.
Darren demonstrated using ELIZA: in the first example a conversation about a problem:
and in the second example the pattern responses built into ELIZA:
Brian had bought some ESP32 microcontrollers which support Bluetooth and Wi-Fi and showed how he had added nodes from the Node-RED website to undertake calculations using them:
He also drew attention to the Tailscale VPN service which uses WireGuard and is free for one personal user with up to 20 devices and the Mi Box streaming media player, noting however its limited language support.
Bernard shared a photograph of a young hedgehog he had encountered out in the road:
David shared the digital thermometer stands he had created with his 3D printer.
Past Meetings