هذه الصفحة متاحة أيضًا باللغة العربية.
التبديل إلى العربيةEsta página también está disponible en español.
Cambiar a españolCette page est également disponible en français.
Passer en françaisBu sayfa Türkçe olarak da mevcut.
Türkçe sürüme geçArtificial intelligence research, applications, data infrastructure and verified corporate announcements. Page 3.
122 stories
BusinessEvent / data period:
The paper introduces a type system for describing clock-sensitive communication protocols between hardware components. It aims to identify mismatches that can lead to unsynchronized exchanges or stalled progress when separately designed components are combined.
BusinessEvent / data period:
The agreements cover training, workshops and founder support for companies adopting agentic AI.
BusinessEvent / data period:
The company said it is the first telecom operator in the region to receive the accreditation after an independent assessment.
BusinessEvent / data period:
Researchers propose UnO-NeSy, a unified ferroelectric field-effect transistor in-memory architecture for neuro-symbolic AI. The paper targets the memory demands of combining neural perception with symbolic reasoning and hyperdimensional computing.
BusinessEvent / data period:
The paper studies a hybrid architecture that pairs GPUs with processing-near-memory resources and shared DRAM for large language model inference. It focuses on prefill and decode phases, which stress compute and memory bandwidth differently when serving requests.
BusinessEvent / data period:
A preprint describes an error model based on read/write tests from a low-power ferroelectric RAM chip carried on a CubeSat in low Earth orbit. The authors say observed bit errors often formed regular spatial patterns, which their generative method learned.
BusinessEvent / data period:
MiniMax announced the M3.1 Flash model for agentic tasks, coding, computer use and cybersecurity. The company says online inference is nearly three times faster and usage costs are 70% lower than its comparison baseline.
BusinessEvent / data period:
A preprint presents SIXPPAC, a mixed-signal machine-learning engine for real-time integrated sensing and communication in wireless baseband. Its design combines two successive-approximation ADCs with a six-layer digital binary-weight neural network.
BusinessEvent / data period:
The preprint studies local serving of smartphone graphical-interface agents using hybrid linear-attention models on mobile neural processing units. Such agents repeatedly inspect a screen and choose an action, so inference delay can accumulate across steps.
BusinessEvent / data period:
A preprint examines combinations of shared scales, weights and activations in MX-like low-precision formats for large language model inference. The researchers jointly explore model accuracy and estimated silicon trade-offs when integrating a tensor core with an edge neural processing unit.
BusinessEvent / data period:
The study describes a generic workflow for developing, deploying and orchestrating AI on Earth-observation satellites. Onboard processing can turn large image collections into smaller outputs such as masks or detections before data are downlinked.
BusinessEvent / data period:
The authors describe an optical transmitter built around thin-film lithium tantalate, aimed at moving data between computing chips with less reliance on power-hungry electrical links. The design combines multiple wavelengths to raise channel capacity while limiting the number of lasers and fibres.
A multilingual perspective on the global economy.
