Catégorie Back-End (46 / 232)
Adelia
Tool offering assessments, white papers and other best practices to verify the ethics of your approaches
Adversarial robustness toolbox
Python library for machine learning security. ART provides tools that enable developers and researchers to defend and evaluate learning (ML) models and applications against adverse threats. ART supports all popular machine learning frameworks (TensorFlow, Keras, PyTorch, MXNet, scikit-learn, XGBoost, LightGBM, CatBoost, GPy, etc.), all data types (images, tables, audio, video , etc.) and machine learning of tasks (classification, object detection, voice recognition, generation, certification, etc.).
Aequitas
Open source bias auditing framework for machine learning developers, analysts and policymakers to audit machine learning models for discrimination and bias, and make informed and fair decisions regarding the development and the deployment of predictive risk assessment tools.
Agaro
Agaro is a solution running on ServiceNow to help you measure and reduce your carbon impact and your electronic waste
#Footprint #Measurement #Service Now
Ai fairness 360 (aif360)
Open-source library of techniques for detecting and mitigating biases in machine learning models throughout the lifecycle of AI applications. The AI Fairness 360 package is available in Python and R.
Cottage
Cabanon is an experimental visualization project exploring issues related to privacy and anonymization techniques by visually comparing the results obtained by anonymizing a dataset with different anonymization methods.
Carbon emissions tracker
Calculates carbon emissions based on battery usage, car travel, and other factors. The information is displayed to the user, promoting environmental awareness.
Carbontracker
Carbontracker is a tool for monitoring and predicting the energy consumption and carbon footprint of learning models (AI)
Code carbon
Estimate and monitor carbon emissions from the calculation, quantify and analyze their impact.Allows you to calculate and reduce CO2 emissions from your IT
Cookieviz
Cookieviz is a dataviz tool offered by the CNIL which allows you to monitor in real time the tracking of your navigation
Cpu energy meter
CPU Energy Meter is a Linux tool that allows you to monitor the energy consumption of Intel processors with fine temporal granularities (a few tens of milliseconds). Power monitoring is available for the following power domains: - by package domain (CPU socket) - by core domain (all processor cores in a package) - by uncore domain (uncore components, by example, integrated graphics on client processors) - by memory node (local memory of a package, server processors only) - by platform (all devices on the platform that are powered by an integrated power mechanism , for example, processor cores, SOC, memory, add-ons or peripherals)
Energy meter logger
Records energy data on a Raspberry Pi and allows you to plot graphs of your energy consumption.
Energybox
EnergyBox is a parameterized tool that allows precise studies of communication energy in the user, using real traffic traces as input. Currently, it focuses on the most popular wireless technologies (3G and WiFi) and emulates the communication energy footprint of application data at the user device level. EnergyBox performs iterative simulation based on the trace.
Energy usage
Python package that provides a function to estimate the energy consumption and associated carbon emissions of another function. Emissions are calculated based on the user's location via the GeoJS API and energy mix data from that location (sources: US EIA and eGRID for the year 2016).
Ethical os toolkit
Toolkit incorporating the following tools: Checklist of 8 risk areas to help you identify the most critical emerging areas of risk and social harm for your team, to start considering now; 14 scenarios to spark conversation and expand your imagination about the long-term impacts of the technology you are developing today; 7 forward-looking strategies to help you take ethical action today.
Ethics & algorithms toolkit
Practical toolkit to help cities and territories understand the implications of using an algorithm, clearly articulate potential risks and identify ways to mitigate them.
Experiment-impact-tracker
Method to track energy consumption, carbon emissions, and calculate your system usage. Currently, on Linux systems with Intel chips (which support RAPL or powergadget interfaces) and NVIDIA GPUs, we record: CPU and GPU power consumption, hardware information, python package versions, information on estimated carbon emissions, etc.
Green algorithms
Small online tool allowing you to identify how green are your calculations?
#Machine Learning #Carbon Footprint #AI #Algorithm #Calculation
Joulemeter
Joulemeter is an interesting application created by Microsoft that shows you the power consumption of each part of your computer. After analyzing your computer, it will calculate the power consumption of each element: motherboard, HD screen, etc. Through this analysis, the program can tell you the power consumption of each element in real time at any time. Joulemeter also records statistics of your consumption so that you can consult them in the future.
Deda (data ethics decision aid)
It is a “toolbox” which helps to map ethical issues in projects involving data processing. This tool was developed, in collaboration with data “practitioners”, by the Utrecht Data School and the University of Utrecht. It allows companies to analyze the growing challenges they are facing.
Ml co2 impact
Calculation of the carbon impact, only for the specific use case of machine learning.
Mlco2
This calculator will give you 2 figures: the gross carbon emissions produced and the approximate offset carbon emissions. The latter number depends on the grid used by the cloud provider and we are open to updating our estimates if anything seems inaccurate or out of date.
#Machine Learning #Carbon Footprint #IA
Power meter
This project includes two different tools: power_meter to record multiple power related sensors as a csv file and make_plot.py to generate power consumption graphs.
Powerapi
PowerAPI is a middleware toolkit for creating software-defined power meters. Software-defined power meters are configurable software libraries that can estimate software power consumption in real time. PowerAPI supports the acquisition of raw metrics from a wide variety of sensors (e.g. physical meters, processor interfaces, hardware counters, OS counters) and the provision of power consumption through different channels (including file system, network, web, graphics). As a middleware toolkit, PowerAPI offers the ability to assemble “à la carte” power meters to meet user needs.
Powertop
PowerTOP is a Linux tool used to diagnose power consumption and power management issues. In addition to being a diagnostic tool, PowerTOP also has an interactive mode that you can use to experiment with various power management settings, in cases where the Linux distribution has not enabled these settings.
Ptopw
pTopW estimates the energy consumption of each process and component in each state. User can change these parameters according to their own platform to get more precise result.
Tfjs-models
This repository hosts a set of pre-trained models that have been ported to TensorFlow.js. The templates are hosted on NPM and unpkg so they can be used in any out-of-the-box project. They can be used directly or in a transfer learning setting with TensorFlow.js.
Transalgo
It is a scientific platform for the development of transparency and accountability of algorithms and data. Transalgo is a project led by INRIA to develop methods for verifying whether a decision rendered by algorithms is ethical.
Cloud carbon footprint
Cloud Carbon Footprint is an open source tool that offers a methodology and tools to measure, monitor and reduce your carbon emissions in the cloud. This approach integrates "Etsy's Cloud Jewels" Cloud kWh Estimate, to convert cloud usage into estimated energy consumption and carbon emissions.
#Cloud #Infrastructure #Footprint #CO2 #carbon
Globaldcanalysis: global data center energy analysis model
The Excel file contains the Global Data Center Energy Analysis Model, including model structure, discussion of uncertainties, and all data entries, results, methodological notes, figures and sources. (Masanet, E., Shehabi, A., Lei, N., Smith, S., and JG Koomey (2020). "Recalibrating global data center energy use estimates." Science, Vol 367, Iss 6481.)
Cloud foundry footprint
Python script that estimates the carbon footprint of all Cloud Foundry (CF) applications in an organization. The script uses basic "cf cli" commands to count the number of running containers per space. This is then multiplied by an estimated average energy consumption
#Cloud #Infrastructure #Footprint #CO2 #carbon
Pyjoules
A software toolkit for measuring the energy footprint of a host machine throughout the execution of a piece of Python code.
Likwid
Likwid is an easy-to-install and use command-line application tool suite for performance-oriented programmers. It works for Intel and AMD processors on Linux operating system.
Powersim
Power / energy simulation in SystemC (not to be confused with VHDL). Powersim is a SystemC class library for calculating the power and energy consumption of hardware described at the system level. To this end, C ++ operators are monitored and different energy models can be used for each type of data.
#Simulation #C ++ #Energy #Power
Ethereum-nft-activity
How much energy does it take to power popular CryptoArt platforms backed by Ethereum? And what emissions are associated with this energy consumption?
#Ether #CryptoArt #Footprint #CO2 #Cloud
Ethereum energy consumption
An approach for the estimation of the energy consumption per day of Ethereum which gives birth to a second carbon footprint estimate project ... but some will be interested by this only energy approach
#Ether #Energy #CryptoArt #Cloud
Data co2 calculator v4
Excel sheet allowing the evaluation of the carbon footprint of data in the different phases of their life cycle