I would characterize this collection as "Digital Neo-Expressionism meets AI Brutalism." This series of research covers rejects the typical "clean and corporate" tech aesthetic in favor of something far more visceral, complex, and intellectually demanding.

The work functions on a principle of Maximalist Layering. We aren't just presenting information; we are visualizing the "black box" of artificial intelligence—the messy, overlapping, and non-linear way that neural networks process data. By using glitch aesthetics and dense digital collages, the covers mirror the complexity of the research papers they represent.
- MemoryFormer: Minimize Transformer Computation by Removing Fully-Connected Layers
- Can LLMs Generate Novel Research Ideas?

- LIMO: Learning Internal Models with Contrastive Oscillatory Recurrence

- A Survey of Human-in-the-loop for Machine Learning


















     


©2022 Ebenz Augustave