Lukas Rambold ramboldio

BeeLearning 🐝

Idea

Do image classification like bees do it. Find a (simple) electric curcuit that can classify a event in an analog input video signal. Train on simulated harware that gives us the gradients.

Gains

dirt cheap to produce harware for
1000x electricity saving for classification

author	title	date	sitemap
Lukas Rambold	Public Sector IT: The Good Parts	02-01-2021	false

By founding a startup in public sector IT I wanted to contribute my software engineering skills where there is a lot left to do and where one can directly contribute to society. After working through that and witnessing the deep systemic issues in the public sector usually are not engineering issues, it is hard not to get discouraged and doubting the long-term survival of democratic governments. Here, I want to hightlight projects (free of criticism and hopefully independent of how much marketing each project did for itself) and system that work well, give hope and should inspire imitation:

Stadt Bühl offer Jitsi Service to schools and other institutions with self-hosted infrastructure for free. No cloud or Microsoft required. Link

Wer wird die Existenzfrage eGovernment beantworten?

Wenige werden bestreiten, dass ein funktionierender Staat im Jahr 2050 noch mit Papierakten hantiert oder mit Stempel und Unterschrift Dokumente bestätigt.

Deshalb ist es weniger ein Frage, ob der Staat der Zukunft ein digitaler sein wird, sondern vielmehr, wer diese Entwicklung antreiben wird. Meine Favoriten: Die bestehenden staatlichen Institutionen, die Zivilgesellschaft/OpenSource Communities oder die Digitalwirtschaft/BigTech.

Dazu illustriere ich, wie die nächsten 10 Jahre, in denen diese Frage entschieden wird in den verschiedenen Szenarien verlaufen können.

	#!/bin/python3

	"""
	opens links in link file formats on all operating systems
	currently supported
	Apple's .webloc & Dropbox' .web

	© Lukas Rambold <[email protected]>, 2022
	MIT License
	"""

	# You will find a step-by-step guide to this example in the docs and the
	# corresponding jupyter notebook on our github repository.

	using DifferentialEquations
	using LightGraphs
	using NetworkDynamics
	using Plots
	using LinearAlgebra
	using DiffEqFlux, DiffEqSensitivity, Flux, OrdinaryDiffEq, Zygote

	<?xml version="1.0" encoding="UTF-8"?>
	<OpenModelicaModelDescription
	xmlns:exp="https://svn.jmodelica.org/trunk/XML/daeExpressions.xsd"
	xmlns:equ="https://svn.jmodelica.org/trunk/XML/daeEquations.xsd"
	xmlns:fun="https://svn.jmodelica.org/trunk/XML/daeFunctions.xsd"
	xmlns:opt="https://svn.jmodelica.org/trunk/XML/daeOptimization.xsd"
	xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
	fmiVersion="1.0"
	modelName="SimpleSeesaw"
	modelIdentifier="SimpleSeesaw"

	====================[ Build \| hyriseCalibration \| Debug ]=======================
	/snap/clion/99/bin/cmake/linux/bin/cmake --build /home/lukas/Projects/hyrise/cmake-build-debug --target hyriseCalibration -- -j 2
	[ 0%] Built target lz4
	[ 0%] Built target resourceLink
	[ 6%] Built target zstd
	[ 6%] Built target cqf
	[ 8%] Built target sqlparser
	[ 10%] Built target libjemalloc-build
	[ 30%] Built target tpcds_dbgen
	[ 32%] Built target hpinuma_base_s

	### Keybase proof

	I hereby claim:

	* I am ramboldio on github.
	* I am ramboldio (https://keybase.io/ramboldio) on keybase.
	* I have a public key whose fingerprint is BEEA EE8C 9D7E 52E1 19B6 E936 D850 90BA 05BF 7AD0

	To claim this, I am signing this object:

	You need to import lodash which allows you to deep copy objects and arrays easily.
	You can also write your own deep copy code of course. :D


	JS
	--------------------------------------------------

	_forceUpdate : function(property)
	// set a deep-copy of product as product
	this.set(property, _.clone(this.get(property)));

	license: gpl-3.0
	height: 600