Freelancer profile translated to English.
Back to original languageAbout Benoît
Engineering graduate, specialist in Explainable Machine Learning; I assist you in extracting relevant information from your data, while detailing the reasoning deployed by the algorithms.
Why Explainable Machine Learning (MLE) rather than simple Machine Learning (ML)?
In ML, when a prediction is requested, there is a score between 0 and 1 for an propensity to an event; typically a purchase probability. This number remains opaque. By delving into the internal parameters of the model, an illustration of the intelligence deployed by the algorithm can be extracted, and used to inform the business. This is the object of MLE.
MLE brings in addition to ML:
- Direction, Strategy: a data-driven, fine, and exhaustive mapping of customer segments; the opening of markets that pass under the signal;
- Data scientists: faster debugging, a focus on explanation
- Marketing, Sales: messages more finely adapted to customers
- Ethics, Audits: knowing the biases hidden in the data
- For the company: data-driven v2.0: Culture of relevant signal, linking the revenue generated to the power of information from an individual information
This set of techniques, also called eXplainable Artificial Intelligence (XAI) is a burgeoning subject at the moment. The American agency DARPA has a program on this theme ("Explainable Artificial Intelligence" - darpa.mil/program/explainable-artificial-intelligence). The Villani mission devoted an entire chapter to it (06 "Opening the black boxes of AI").
Missions:
- You want to enhance your data, I help you set up a predictive pipeline
- You have an existing ML pipeline, I help you extract the reasoning deployed by your algorithms
Technologies:
- RuleFit, Random Forest, Word2Vec, PCA, ALS, t-SNE, LSH, ROC
- Scikit-LearnSpark, Weka, Databricks, BigQuery, Hive
- Postgres, MySQL, Oracle
- AWS, Linux, Maven, Git
- Python, Java, Scala, CAML, Elm
- Spring, Primefaces, d3.js
English
Native or bilingual
Chinese
Basic
Spanish
Conversational
Can work on-site
Paris (up to 10km)
Experience
- Participation à l'Explainable Machine Learning ChallengeCompetition supervised by Google, FICO, and under the aegis of the NIPS conferenceSOFTWARE PUBLISHINGJuly 2018 - August 2018 (2 months)Challenge site:• Improvement of a standard algorithm (RuleFit) to make its explanations more succinct• Creation of an intuitive, explanatory map of decision modes• Available in open-source: https://github.com/benoitparis/explainable-challengeTechnologies: Scikit-Learn, Python, t-SNE, Lasso, Jupyter
- Benoit Paris ConsultingMachine Learning EngineerCONSULTING AND AUDITSJune 2017 - Today (9 years)Paris, France
- Predictive marketing pipeline, evaluation of predictions
- Predictive model explanation solution (Proprietary Algorithm - Spark)
- Proof of Concept: Listing of vector analogies by indexing in high dimensions
- Clémence ConsultingData Management EngineerCONSULTING AND AUDITSDecember 2012 - December 2022 (10 years)
- Batch architecture for multi-channel marketing campaigns, administration and back office tools, BI
- BDD administration, data quality, analyses, reconciliations, enrichments, datamining, data flows
Recommendations
Be the first to recommend Benoît
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Engineering degree - Management Information Systems & e-CommerceEcole Centrale de Lille2010