Goto

Collaborating Authors

 Telecommunications


Microsoft's Copilot AI goes head-to-head with China's DeepSeek in Africa

The Japan Times

Microsoft's Copilot AI goes head-to-head with China's DeepSeek in Africa Microsoft is investing 5.4 billion South African rand ($330 million) to expand its cloud and AI capacity in the country by the end of next year, and it also has plans to build a geothermal-powered data center in Kenya. Microsoft is making a push for more Africans to adopt its artificial-intelligence tools as the U.S. technology giant competes with China's DeepSeek for customers from the world's youngest and fastest-growing population. The Redmond, Washington-based company plans to train 3 million Africans on its AI technology this year, in partnership with schools, universities and other institutions, with a focus on South Africa, Kenya, Nigeria and Morocco. It's also partnered with MTN Group, Africa's biggest telecommunications firm, to sell the Microsoft 365 suit of apps together with its Copilot digital assistant to its 300 million subscribers. The Microsoft Elevate training initiative aims to make sure cost is not a barrier to building AI literacy at scale," Middle East and Africa President Naim Yazbeck said in an interview. Chinese technology is active in Africa and our job is to compete."



A Simple yet Scalable Granger Causal Structural Learning Approach for Topological Event Sequences

Neural Information Processing Systems

Such causal graphs delineate the relations among alarms and can significantly aid engineers in identifying and rectifying faults. However, existing methods either ignore the topological relationships among devices or suffer from relatively low scalability and efficiency, failing to deliver high-quality responses in a timely manner.



Andrey Kuzmin, Markus Nagel, Mart van Baalen, Arash Behboodi, Tijmen Blankevoort Qualcomm AI Research

Neural Information Processing Systems

In this paper, we set out to answer the question on which is better: neural network quantization or pruning? By answering this question, we hope to inform design decisions made on neural network hardware going forward. We provide an extensive comparison between the two techniques for compressing deep neural networks.




SoftBank swings to profit on valuation boost from OpenAI bet

The Japan Times

SoftBank CEO Masayoshi Son (left) and OpenAI CEO Sam Altman attend an event in Tokyo in February 2025. SoftBank's investment gain on OpenAI stood at an estimated $19.8 billion as of December. SoftBank Group sprang back to a quarterly profit after investment gains from OpenAI neared $20 billion, a promising start for one of CEO Masayoshi Son's signature gambles alongside ByteDance and Alibaba Group Holding. The Tokyo-based company has invested about $34.6 billion in OpenAI, accumulating an 11% stake as of December, and has been in talks to invest as much as $30 billion more in a round that would value the startup at about $750 billion to $830 billion. As of December, SoftBank's investment gain on OpenAI stood at $19.8 billion, the company said Thursday.


SoftBank swings to profit on valuation boost from OpenAI bet

The Japan Times

SoftBank CEO Masayoshi Son and OpenAI CEO Sam Altman attend an event in Tokyo in February 2025. SoftBank's investment gain on OpenAI stood at an estimated $19.8 billion as of December. SoftBank Group sprang back to a quarterly profit after Masayoshi Son's bet on OpenAI paid off in valuation gains, cementing the Japanese company's position as an investment proxy for the ChatGPT creator. The Tokyo-based company has invested more than $30 billion (¥4.58 trillion) in OpenAI, accumulating an 11% stake as of December, and has been in talks to invest as much as $30 billion more in a round that would value the startup at about $750 billion to $830 billion. As of December, SoftBank's investment gain on OpenAI stood at an estimated $19.8 billion, the company said Thursday.


FastRoutingunder Uncertainty: AdaptiveLearning inCongestionGameswithExponentialWeights

Neural Information Processing Systems

We examine an adaptive learning framework for nonatomic congestion games where the players' cost functions may be subject to exogenous fluctuations (e.g., due to disturbances in the network, variations in the traffic going through a link, etc.).