The latest publications in this offer are:
Our ability to manipulate individual particles over the past 20 years has enabled a new quantum revolution. This new revolution, known as the “second quantum revolution,” is characterized by the ability to exploit quantum effects for engineering purposes.
This paper presents an overview of the state of the art in natural language processing, exploring one specific computational architecture, the Transformer model, which plays a central role in a wide range of applications. This architecture condenses many advances in neural learning methods and can be exploited in many ways: to learn representations for linguistic entities; to generate coherent utterances and answer questions; to perform utterance transformations, an illustration being machine translation; to serve as backbone of versatile chatbots, able to answer queries and to perform, at will, a large variety of tasks. These facets of the architecture will be successively presented, which will also allow us to discuss its limitations.
Operationalizing AI ethics is a major challenge for organizations. The Amazon, COMPAS and Obermeyer cases illustrate how AI systems can produce systemic discrimination despite the absence of malicious intent. This article proposes an operational framework for transforming ethical principles into concrete risk and impact management practices, based on the ISO/IEC 42001:2023 standard. We first show how this standard aligns with GDPR requirements in Europe and Law 25 in Quebec. We then detail the integration of risk and impact management at different stages of the AI systems lifecycle. Finally, we present operational tools (lifecycle × risks × controls matrix, indicators, organization) and a progressive roadmap for implementation.
In 2008 Satoshi Nakamoto defined a new model of currencies. The issuance and management of such a currency take place on a peer-to-peer network without central control. Bitcoin is the first cryptographic currency created on this model. It exists since January 2009.As the 7000 other crypto-currencies of the same type created in its wake, it works through what is called a blockchain. This shared file and collectively controlled file by a peer-to-peer network can be used for many other purposes. Many applications are developed and implemented gradually, especially in the world of banking and finance.
Nowadays, testing complex systems is not easy and the time allocated to validation teams is often considered too short. This article explains how to properly define, plan, track, and optimize the testing of complex systems. It falls within a systems engineering context. It describes the processes associated with the test and gives best practices. It proposes a test typology that facilitates the development of the test strategy. It addresses the issue of documentation and test management. It presents various optimization criteria to minimize test costs and risks during and at the end of the test.
This paper discusses the European regulation on artificial intelligence systems, the AI Act. This regulation addresses the principles of good management of an artificial intelligence system using a risk-based approach and provides related compliance and control procedures. The article provides a series of use cases and feedback to facilitate the understanding and integration of the AI Act within organizations.
Cloud Computing allows businesses and individuals to deploy data and applications among external infrastructures, leased when needed, and released when useless. It is possible thanks to the ubiquitous network and to virtual machines techniques, made available by the performances of modern microprocessors. This dissemination of data lessens data losses risks, but creates new risks related to privacy and data corruption. And there are "black clouds", used by cybercrime organisations.
Facial recognition, which relies on common methods of processing faces but whose uses differ depending on the legal framework, is a major component of artificial intelligence. This article offers a comparative analysis of its deployment in China, the United States and France. In China, it fuels mobile payments and social credit, promoting widespread surveillance despite the PIPL. In the United States, regulatory debates remain heated, as illustrated by the Trump administration's repeal of the Biden executive order. In France, the executive branch supports its expansion, despite reservations from the Defender of Rights. The central question remains: is a compromise between security and personal data protection possible?
Artificial intelligence - AI - methods and big data processing are essential in process and product engineering, a complex and interdisciplinary science. Multilinear regression and principal component analysis are already common, but supervised, unsupervised, and combinatorial learning methods are also used, with new approaches constantly emerging. These techniques simplify the development of models and solve complex problems in various sectors: chemistry, food, etc. This article examines AI methods, their relevance, and their applications, while also addressing the challenges of physical interpretability and extrapolation.
Information retrieval covers many applications ranging from document retrieval using Boolean queries to the generation and extraction of precise answers to questions in natural language. It applies to text, images, and audio, and can be interactive, in the form of dialogues with a conversational agent. This article focuses on the crossing of information retrieval and generative AI, known as augmented (response) generation through information retrieval (RAG). RAG assists in the generation of responses from a large language model and information sources that may be private. Large language models and RAG architectures are presented (agentic RAG, GraphRAG, etc.), as are the many strategies that can be followed.
Access to a comprehensive database of articles
Rely on the ultimate technical and scientific knowledge database
Contents written and validated by leading experts to ensure reliability and expertise
Our experts committeeQuizzes, medias, tables, formulas, videos, practical paths, etc...
Enhanced reading experience
Reference articles available in English and French
Updates available in both languages
|
Single-user
2366 € excl. VAT
|
||
| Knowledge database | ||
| Number of licenses | Depending on company size | 1 |
| Languages | French and/or English | French and/or English |
| Access to enhanced reference articles | Unlimited | Unlimited |
| Access to archives | ||
| Publication updates and monitoring tools | ||
