Technology
Researchers Develop Algorithm for Optimal Decision Making Under Heavy-tailed Noisy Rewards
By Adedapo Adesanya
The exploration algorithms for stochastic multi-armed bandits (MABs)–sequential decision-making problems under uncertain environments–typically assume light-tailed distributions for reward noises.
However, real-world datasets often show heavy-tailed noise. In light of this, researchers from Korea propose an algorithm that can achieve minimax optimality (minimum loss under maximum loss scenario) with minimal prior information.
Superior to existing algorithms, the new algorithm has potential applications in autonomous trading and personalized recommendation systems.
In data science, researchers typically deal with data that contain noisy observations. An important problem explored by data scientists in this context is the problem of sequential decision-making. This is commonly known as a “stochastic multi-armed bandit” or (stochastic MAB).
Here, an intelligent agent sequentially explores and selects actions based on noisy rewards under an uncertain environment. Its goal is to minimize cumulative regret–the difference between the maximum reward and the expected reward of selected actions. A smaller regret implies more efficient decision-making.
Most existing studies on stochastic MABs have performed regret analysis under the assumption that the reward noise follows a light-tailed distribution. However, many real-world datasets, in fact, show a heavy-tailed noise distribution.
These include user behavioural pattern data used for developing personalized recommendation systems, stock price data for automatic transaction development, and sensor data for autonomous driving.
In a recent study, Assistant Professor Kyungjae Lee of Chung-Ang University and Assistant Professor Sungbin Lim of the Ulsan Institute of Science and Technology, both in Korea, addressed this issue. In their theoretical analysis, they proved that the existing algorithms for stochastic MABs were sub-optimal for heavy-tailed rewards.
More specifically, the methods employed in these algorithms–robust upper confidence bound (UCB) and adaptively perturbed exploration (APE) with unbounded perturbation–do not guarantee a minimax (minimization of maximum possible loss) optimality.
“Based on this analysis, minimax optimal robust (MR) UCB and APE methods have been proposed. MR-UCB utilizes a tighter confidence bound of robust mean estimators, and MR-APE is its randomized version. It employs bounded perturbation whose scale follows the modified confidence bound in MR-UCB,” explains Dr Lee, speaking of their work, which was published in the IEEE Transactions on Neural Networks and Learning Systems on 14 September 2022.
The researchers next derived gap-dependent and independent upper bounds of the cumulative regret. For both the proposed methods, the latter value matches the lower bound under the heavy-tailed noise assumption, thereby achieving minimax optimality.
Further, the new methods require minimal prior information and depend only on the maximum order of the bounded moment of rewards. In contrast, the existing algorithms require the upper bound of this moment a priori–information that may not be accessible in many real-world problems.
Having established their theoretical framework, the researchers tested their methods by performing simulations under Pareto and Fréchet noises. They found that MR-UCB consistently outperformed other exploration methods and was more robust with an increase in the number of actions under heavy-tailed noise.
Further, the duo verified their approach for real-world data using a cryptocurrency dataset, showing that MR-UCB and MR-APE were beneficial–minimax optimal regret bounds and minimal prior knowledge–in tackling heavy-tailed synthetic and real-world stochastic MAB problems.
“Being vulnerable to heavy-tailed noise, the existing MAB algorithms show poor performance in modelling stock data. They fail to predict big hikes or sudden drops in stock prices, causing huge losses. In contrast, MR-APE can be used in autonomous trading systems with stable expected returns through stock investment,” comments Dr Lee, discussing the potential applications of the present work.
“Additionally, it can be applied to personalized recommendation systems since behavioural data shows heavy-tailed noise. With better predictions of individual behaviour, it is possible to provide better recommendations than conventional methods, which can maximize the advertising revenue,” he concludes.
By Modupe Gbadeyanka
A leading telecommunications firm, Airtel Nigeria, has reaffirmed its long-term commitment to strengthening the country’s digital infrastructure and data access to bridge gaps in connectivity and unlock new opportunities in the country.
The company gave this reassurance during a recent inspection tour of its ongoing Nxtra Data Centre at Eko Atlantic, Lagos.
The data centre is being established to deliver hyperscale and edge facilities across key African markets. With a load of 38 Megawatts, the Lagos facility is expected to serve as a major hub for data hosting, cloud services, content distribution, artificial intelligence, and enterprise solutions in West Africa.
“This Nxtra Data Centre in Lagos represents a critical part of our long-term vision for Nigeria’s digital ecosystem. Today’s visit allows us to review progress, engage our stakeholders, and ensure that our infrastructure investments continue to meet global standards and local needs.
“This data centre will deliver critical high multi megawatt capacity in line with hyperscale customers and enable high density environment. We are putting the infra to bring the cloud to Nigeria,” the chief executive of Airtel Africa Plc, Mr Yashnath Issur, said.
Also commenting, the chief executive of Airtel Nigeria, Mr Dinesh Balsingh, said, “Since the announcement of this project, our focus has been on building a world-class facility that supports Africa’s digital transformation agenda.
“We are encouraged by the progress recorded so far and remain committed to delivering a secure, energy-efficient, and future-ready data centre for Nigeria,” reiterating that the data centre is progressing steadily towards the previously announced 2028 go live date.
On his part, the chairman of Eko Atlantic, Mr Gabbi Massoud, disclosed that, “Eko Atlantic as a city with high quality infrastructure will contribute positively to boost the economy of Nigeria and is a perfect place for the development of the digital infrastructure of Nigeria.
“The Nxtra data centre reflects the calibre of projects we seek to attract — long-term, technology-driven investments built to the highest global standards.
“Today’s visit affirms the rigour of the planning and execution process by Nxtra, and the commitment of Eko Atlantic to facilitate and promote the Nigeria’s evolving digital ecosystem.”
By Modupe Gbadeyanka
A speech dataset designed to catalyze research and build more inclusive Artificial Intelligence (AI) technologies has been launched by Google in partnership with a consortium of leading African research institutions, which are mainly universities.
The main universities involved in the project known as WAXAL are Makerere University in Uganda, the University of Ghana, and Digital Umuganda in Rwanda.
A statement from Google on Monday said the dataset bridges a critical digital divide for over 100 million speakers by providing foundational data for 21 sub-Saharan African languages, including Hausa, Luganda, Yoruba, and Acholi.
While voice-enabled technologies have become common in much of the world, a profound scarcity of high-quality speech data has prevented their development for most of Africa’s over 2,000 languages. This has excluded hundreds of millions of people from accessing technology in their native tongues.
The WAXAL dataset was created to directly address this gap. Developed over three years with funding from Google, the project features 1,250 hours of transcribed, natural speech, and Over 20 hours of high-quality, studio recordings designed for building high-fidelity synthetic voices.
The WAXAL dataset, which is available starting today, covers Acholi, Akan, Dagaare, Dagbani, Dholuo, Ewe, Fante, Fulani (Fula), Hausa, Igbo, Ikposo (Kposo), Kikuyu, Lingala, Luganda, Malagasy, Masaaba, Nyankole, Rukiga, Shona, Soga (Lusoga), Swahili, and Yoruba.
Commenting on the development, the Head of Google Research for Africa, Ms Aisha Walcott-Bryantt, said, “The ultimate impact of WAXAL is the empowerment of people in Africa.
“This dataset provides the critical foundation for students, researchers, and entrepreneurs to build technology on their own terms, in their own languages, finally reaching over 100 million people.
“We look forward to seeing African innovators use this data to create everything from new educational tools to voice-enabled services that create tangible economic opportunities across the continent.”
Also commenting, a Senior Lecturer at Makerere University’s School of Computing and Information Technology, Ms Joyce Nakatumba-Nabende, said, “For AI to have a real impact in Africa, it must speak our languages and understand our contexts.
“The WAXAL dataset gives our researchers the high-quality data they need to build speech technologies that reflect our unique communities. In Uganda, it has already strengthened our local research capacity and supported new student and faculty-led projects.”
An Associate Professor at the University of Ghana, Mr Isaac Wiafe, said, “For us at the University of Ghana, WAXAL’s impact goes beyond the data itself. It has empowered us to build our own language resources and train a new generation of AI researchers.
“Over 7,000 volunteers joined us because they wanted their voices and languages to belong in the digital future.
“Today, that collective effort has sparked an ecosystem of innovation in fields like health, education, and agriculture. This proves that when the data exists, possibility expands everywhere.”
By Adedapo Adesanya
The Nigeria Data Protection Commission (NDPC) has disclosed that the country’s data protection ecosystem has grown to N16.2 billion within just two years of formal regulation.
The disclosure was made by the chief executive of the data regulating agency, Mr Vincent Olatunji, during a media workshop and capacity-building engagement held in Lagos recently.
He further said the growth reflects rising enforcement, compliance activity, and increasing confidence in Nigeria’s digital governance framework, even though the NDPC was not designed as a revenue-generating agency.
Mr Olatunji explained that regulatory compliance fees and enforcement actions under the Nigeria Data Protection Act (NDPA), 2023, have created significant economic value while also contributing to government revenue and job creation across the country, noting that regulatory fees and sanctions after investigations have contributed over N16.2 billion to federal revenue while supporting an estimated 23,000 jobs nationwide.
“These investigations have resulted in 11 major enforcement actions, including significant financial penalties and corrective directives.”
“The message is clear: violations of data privacy will attract serious consequences, regardless of the size or status of the organisation involved,” Mr Olatunji stated, adding that the commission has concluded 246 investigations into data protection and privacy breaches across multiple sectors, signalling that enforcement will remain central to Nigeria’s data governance strategy.
Business Post reports that NDPC has over the last two years carried some sanctions against some top companies including a N766.2 million fine on MultiChoice Nigeria in July 2025 as well as Fidelity Bank, which was fined N555.8 million in 2024 for processing personal data without informed consent.
The NDPC Commissioner linked the Commission’s enforcement milestones to Nigeria’s broader ambition of building a $1 trillion digital economy.
He stressed that accountability and trust are foundational to digital transformation and long-term investment.
“Privacy enforcement is the foundation of digital confidence. By holding violators accountable, we are safeguarding citizens while creating the secure environment required for innovation, investment and sustainable growth,” he said.
He said the Commission has significantly expanded compliance structures across the economy to support this objective, moving beyond sanctions to system-wide institutional strengthening.
The NDPC has registered 38,677 Data Controllers and Processors of Major Importance, licensed 307 Data Protection Compliance Organisations, and received more than 8,155 Compliance Audit Returns.
In addition, the Commission has issued the General Application and Implementation Directive, which takes effect from September 2025, translated the NDPA into three major Nigerian languages, and launched a multi-sector compliance sweep covering banking, insurance, pensions, and gaming, with 1,348 entities already served with compliance notices.
-
Feature/OPED6 years agoDavos was Different this year
-
Travel/Tourism9 years ago
Lagos Seals Western Lodge Hotel In Ikorodu
-
Showbiz3 years agoEstranged Lover Releases Videos of Empress Njamah Bathing
-
Banking8 years agoSort Codes of GTBank Branches in Nigeria
-
Economy3 years agoSubsidy Removal: CNG at N130 Per Litre Cheaper Than Petrol—IPMAN
-
Banking3 years agoSort Codes of UBA Branches in Nigeria
-
Banking3 years agoFirst Bank Announces Planned Downtime
-
Sports3 years agoHighest Paid Nigerian Footballer – How Much Do Nigerian Footballers Earn