Mokhtar SELLAMI Advisor Science Technology and Innovation / Directorate-General Scientific Research
Artificial Intelligence market worth $190.61 Billion by 2025, and will contribute to the global GDP by nearly 25.5% of about $13 Trillion in the future. To benefit from this advantage and join the worldwide AI national programmes, the minister of higher education and scientific research launched the Algerian strategic plan, on 1-2 december 2019, at Constantine. « We expect results that strengthen the place of artificial intelligence in the lives of our citizens. This requires the promotion of national capacities in the fields of training, education and research, on the one hand, and that artificial intelligence will promote economic development and gives to the socio-economic sector the means to remove the obstacles to the digital transformation underway, on the other hand », said the minister. During this workshop attented by more than 200 stakeholders, conferences and presentations were made by senior experts on the AI strategies over the world, the national capacities (scientific production, human ressources, licences, masters and Ph.D training, main national conférences in AI) and the existing infrastructures (research laboratories, HPC centres, network). The minister express the algerian government commitment to the national plan presented by Prof. Hafid AOURAG, Director-General for Scientific Research and Technological Development. Parallel sessions were organised under the coordination of Prof. GUESSOUM Ahmed (USTHB), and includes 5 tracks :
- Track1 Automated Reasoning and Machine Learning, President: Prof. Hamamache KHEDDOUCI (University of Lyon I) : This part deals with fundamental approaches to representation and resolution problems whether by rules-based approaches (symbolic AI), statistics, or based on machine learning (digital AI). Certainly, in recent years, the different neural network architectures whose Deep Learning attracts efforts of researchers for applications that cover all areas of life human but different other themes are important including more developments in the field of data-based AI hybridization and that which uses levels of reasoning to arrive at an explanation of the results of recommendations, classifications, predictions, etc. ; a hybridization called cognitive technology.
When it comes to digital AI (statistical or machine learning based), also think of approaches to obtain validated sectoral data, which would allow the development of reliable and useful research. Various themes can be mentioned here, without their list being exhaustive (like this will be the case for the other tracks below).
Themes discussed: - Knowledge Management; Fuzzy Logic; Fuzzy Systems; Neural Networks; Deep Learning; Convolutional Neural Networks; Recurrent, - Neural Networks; Convolutional Generative Networks; Adversarial Networks; - Deep Convolutional Generative Adversarial Networks; Capsule Networks; Data Mining, - Genetic Algorithms and Meta-Heuristics
- Track2 : Networking and Infrastructures in the AI Context, President: Prof. Omar BOUSSAID (University of Lyon) : In this track it is a question of covering various aspects of AI and its interactions with themes linked to data transmission networks. The Internet offers immense applications but also, more recently, the concepts of Blockchain, cloud computing, High Performance Computing, Internet of Things (IoT), etc. So it's about thinking about infrastructures that will be able to support training and research in AI as well as the development of applications for all sectors. Security aspects in this context are of course crucial.
Themes discussed: -Internet of Things (Iot); Big Data; Blockchain; Infrastructures and Cloud Computing; High Performance Computing (HPC); Cyber Security; Network ; Optimization; Smart cities; Intelligent Networks, 5G, etc.
-Track3 : Computer Vision, President : Prof. Abdelkader BENYETTOU (University of UST-Oran) : An important area of AI is computer vision. Analysis and recognition of images and videos and hence the field of virtual reality and augmented reality, have very important applications in all fields (medicine, science and technology, tourism, etc.) as well as in training in these various fields.
Themes discussed: - Image Processing; - Image Recognition; - Facial Recognition; - Medical Imaging; - Virtual Reality / Augmented Reality, - Biometrics, etc.
-Track4 : Speech and Language Processing, President : Prof. Mokhtar SELLAMI (University of Annaba) : Automatic Processing of Natural Language (TALN), in text and speech, saw a progressing development, especially with applications related to massive content from the Internet, social networks, mobile applications, etc. It's a domain that can allow the development of applications and startups enough quickly with a certain return on investment (RoI). Let R & D focus on the information retrieval, automatic summary, automatic translation, virtual assistants, or others, a lot can be done in this field, and its related applications.
Themes discussed: Information Retrieval; Text Summarization; Machine Translation; Automatic Speech Recognition; Voice Recognition; Speech Synthesis; Virtual Assistants
Track 5: AI Applications : President : Dr. Brahim BOUZOUIA (Research centre CDTA) : AI is by definition synonymous with applications in all areas of life human. Therefore, it is a question of thinking about R & amp; D while thinking of the socio-national economic priorities.
Themes discussed are: AI in Energy; Industry 4.0; RFID; Intelligent Manufacturing; Productive Maintenance; Automobile Industry; Automotive Industry; Autonomous Vehicles; Robotics; Education; Health; Agriculture; Environment; Transport; Smart Cities; Home Automation; Decision Making; Fintech; Defense;
- Track 6: Legal and Ethical issues of AI : President : Prof. Mustapha KHIATI( University Alger1) : Experts agrees that never a scientific / technological breakthrough has had as much real and future impact as AI. This poses indeed major challenges to all of humanity, philosophical, ethical, legal challenges, and related to people's privacy. It is about defining guiding principles for the development of a clear AI policy that will take into account development of R&D respectful of the noblest human values and more sacred.
Themes discussed are: Privacy; ethics; philosophical challenges; Legal issues.
The participants of the different tracks were invited to discuss the various questions taking as a perspective the general theme as well as their specific themes. Wich agenda 2030 in IA for Algeria ? Which goals to be reached in 2030 in education, in fondamental and applied research, and which impact on the society and the economy. These objectives are to declin for 2023, 2025 and 2030. Key performance indicators (KPIs) for each period are more than welcome.
What are the AI development priorities for the present, in 2025 and 2030? What approaches should be used to achieve these objectives, and through what type of governance (ministry, national council, research center or public industrial and commercial enterprises (EPIC) ?
The current computer park include different types of equipment (workstations, high performance computing center, computer networks, capacities through the cloud), do we need more resources for the development of an industry of software? How to generalize the teaching of AI in the various courses of education in license, master and doctorate to reach the global goal?
Regarding the needs to overcome the gap in society and industry, it was obvious to target some specific questions as how to develop applied AI research to meet the needs of socioeconomic sectors in the fields of energy, education, health, agriculture, environment, industry, transport, administration, governance , finances? How to develop and / or strengthen partnership relations with the socio-economic sector? What funding mechanisms do you suggest?
Finally, participants are invited to reflect on the impact of AI on society and the economy by addressing in particular the aspects related to the popularization of AI in high schools, the public service and the private sector.
These general questions to all tracks are followed by specific one for each track. If we consider track4 in Speech and Language Processing, the participants first worked on the identification of programmes, themes and axes, in research, training and impact (innovation). The industrial applications correspond to techniques for the reading of documents, the production of documents, the human-machine interfaces, the analysis of the voice signal (arabic, tamazigh, management of dialectal languages) and a multi and interdisciplinary component relating to the preservation of cultural heritage. These families of applications are introduced in the following paragraphs.
1) Identification of the different dialects in Algeria (Word and text): The country has at least ten official dialects, the problem is to develop automatic processes for identifying texts spoken or written in Tifinagh, Arabic or Latin characters.
2) Identification of the speaker (on the basis of the spoken word): Develop tools which make it possible to model sign language (combining hand movements and facial expressions in real time) to translate it into a language (Arabic, Tamazight, English, French) written and vice versa. The objective is to facilitate access for deaf people to information and knowledge.
3) Automatic text summary: The automatic summary consists in offering to the user a synthetic and relevant form of an input text, which faithfully and best relates the idea of the overall text.
4) Opinion mining: Opinion mining is a process allowing to know the opinions of users on a given subject and to be used in particular for prediction and decision-making (analysis of feelings, and polarity).
5) Automatic translation: The purpose of automatic translation is to translate spoken or written language from a source language into a target language.
6) Speech recognition (Arabic, Tamazight), particularly in noisy surroundings: This involves transforming a speech signal received by means of a microphone and transcribing it as text. The presence of background noise and variable conditions (environment, etc.) can significantly affect the quality of automatic voice recognition and increases the complexity of the processing.
7) Speech synthesis (text to speech Arabic Tamazight) It consists in reading a text with a synthetic voice, which is a reverse speech to text functionality.
8) Language Processing and Health (aphasia, autism, dyslexia): It is a question of proposing LP tools for better prevention and treatment of various health-related pathologies (autism, dyslexia, medical records, etc.).
9) Automatic handwriting recognition (On / Off line): Handwriting recognition uses shape recognition, to recognize words and sentences existing in handwritten text. A distinction is made between offline recognition and online recognition (privileged use of automatic sorting of postal mail, automatic processing of administrative files, survey forms, or the recording of bank checks).
10) TALN and fraud (detection of plagiarism): Academic fraud is considered as a kind of corruption which can take several forms, including plagiarism, which consists in using the words, ideas or works of others without making a reference explicitly or intentionally merge extracts copied from many sources without reformulation (sale of ready-to-use academic work in cyberspaces and validated by their teachers, etc.). Document forgery is also a defect to be taken care of.
11) Language learning (spell checker) [computer aided language learning based speech]: NLP techniques can be of great help in facilitating language learning through the use of assisted learning systems. computer, or correcting the user's answers to questions related to the grammar of the language, the meaning of words and sentences.
12) TALN and people with specific needs (sign language): Develop tools that allow to model sign language (combining hand movements and facial expressions in real time) to translate it into a language (Arabic, Tamazight , English, French) written and vice versa. The objective is to facilitate access for deaf people to information and knowledge. A challenge on which artificial intelligence programs are still incompetent.
13) Preservation and enhancement of the national cultural heritage (archives, etc.): It is particularly a multidisciplinary field based essentially on the collection, acquisition, restoration, safeguarding, and the provision of researchers of this heritage (data-sets, treasure project of the Arabic language).
Track4 participants : SELLAMI Mokhtar (DGRSDT), ABBAS Mourad (CRTDLA-Modérateur), TLEMSANI Redouane (USTO), Aliane Hassina (CERIST), KHADIR Ahlem Chérifa (CERIST), CHIKHI Salim (Univ Constantine2), HAGANI Iheb Chems El Dine (USTO), AMROUNE Mohamed (Univ-Tebessa), ASBAI Nassim (Univ BBA), HAMOU reda Mohamed (Univ. Saida),BOUMAHDI Fatima (Univ U.Blida1), Amine Abdel Malek (Univ Saida), BENOUDJIT Nabil (Univ. Batna2), BANATIA Mohamed Akram (EMP, Alger), MESSAI Abderraouf (Univ Constantine 1), OUARDI Yakoub Abdelkader (Univ Saida), ZERARI Naima (Univ Batna2), BENSLIMANE Sidi Mohamed (ESI-sidi bel abbas)