Edge deployment of large Vision-Language Models (VLMs) increasingly relies on flash-based weight offloading, where activation sparsification is used to reduce I/O overhead.

Theglobalmodel of differential privacy,which assumes that users trust the data collector, provides strong privacy guarantees and introduces small errors in the output.

Edge deployment of large Vision-Language Models (VLMs) increasingly relies on flash-based weight offloading, where activation sparsification is used to reduce I/O overhead. However, conventional sparsification remains model-centric, selecting neurons solely by activation magnitude and neglecting how access patterns influence flash performance. We present Neuron Chunking, an I/O-efficient sparsification strategy that operates on chunks (i.e., groups of contiguous neurons in memory) and couples neuron importance with storage access cost. The method models I/O latency through a lightweight abstraction of access contiguity and selects chunks with high utility, defined as neuron importance normalized by estimated latency. By aligning sparsification decisions with the underlying storage behavior, Neuron Chunking improves I/O efficiency by up to 4.65x and 5.76x on Jetson Orin Nano and Jetson AGX Orin, respectively.

AMD GPUs offer state-of-the-art compute and memory bandwidth; however, peak performance AMD kernels are written in raw assembly. To address the difficulty of mapping AI algorithms to hardware, recent work proposes C++ embedded and PyTorch-inspired domain-specific languages like ThunderKittens (TK) to simplify high performance AI kernel development on NVIDIA hardware. We explore the extent to which such primitives -- for explicit tile-based programming with optimized memory accesses and fine-grained asynchronous execution across workers -- are NVIDIA-specific or general. We provide the first detailed study of the programming primitives that lead to performant AMD AI kernels, and we encapsulate these insights in the HipKittens (HK) programming framework. We find that tile-based abstractions used in prior DSLs generalize to AMD GPUs, however we need to rethink the algorithms that instantiate these abstractions for AMD. We validate the HK primitives across CDNA3 and CDNA4 AMD platforms. In evaluations, HK kernels compete with AMD's hand-optimized assembly kernels for GEMMs and attention, and consistently outperform compiler baselines. Moreover, assembly is difficult to scale to the breadth of AI workloads; reflecting this, in some settings HK outperforms all available kernel baselines by $1.2-2.4\times$ (e.g., $d=64$ attention, GQA backwards, memory-bound kernels). These findings help pave the way for a single, tile-based software layer for high-performance AI kernels that translates across GPU vendors. HipKittens is released at: https://github.com/HazyResearch/HipKittens.

Detecting malware, especially ransomware, is essential to securing today's interconnected ecosystems, including cloud storage, enterprise file-sharing, and database services. Training high-performing artificial intelligence (AI) detectors requires diverse datasets, which are often distributed across multiple organizations, making centralization necessary. However, centralized learning is often impractical due to security, privacy regulations, data ownership issues, and legal barriers to cross-organizational sharing. Compounding this challenge, ransomware evolves rapidly, demanding models that are both robust and adaptable. In this paper, we evaluate Federated Learning (FL) using the Sherpa.ai FL platform, which enables multiple organizations to collaboratively train a ransomware detection model while keeping raw data local and secure. This paradigm is particularly relevant for cybersecurity companies (including both software and hardware vendors) that deploy ransomware detection or firewall systems across millions of endpoints. In such environments, data cannot be transferred outside the customer's device due to strict security, privacy, or regulatory constraints. Although FL applies broadly to malware threats, we validate the approach using the Ransomware Storage Access Patterns (RanSAP) dataset. Our experiments demonstrate that FL improves ransomware detection accuracy by a relative 9% over server-local models and achieves performance comparable to centralized training. These results indicate that FL offers a scalable, high-performing, and privacy-preserving framework for proactive ransomware detection across organizational and regulatory boundaries.

Caching and prefetching techniques are fundamental to modern computing, serving to bridge the growing performance gap between processors and memory. Traditional prefetching strategies are often limited by their reliance on predefined heuristics or simplified statistical models, which fail to capture the complex, non-linear dependencies in modern data access patterns. This paper introduces a modular framework leveraging Graph Neural Networks (GNNs) to model and predict access patterns within graph-structured data, focusing on web navigation and hierarchical file systems. The toolchain consists of: a route mapper for extracting structural information, a graph constructor for creating graph representations, a walk session generator for simulating user behaviors, and a gnn prefetch module for training and inference. We provide a detailed conceptual analysis showing how GNN-based approaches can outperform conventional methods by learning intricate dependencies. This work offers both theoretical foundations and a practical, replicable pipeline for future research in graph-driven systems optimization.

Sparse Ternary General Matrix-Matrix Multiplication (GEMM) remains under-optimized in existing libraries for Apple Silicon CPUs. We present a Sparse Ternary GEMM kernel optimized specifically for Apple's M-series processors. We propose a set of architecture-aware optimizations, including a novel blocked and interleaved sparse data format to improve memory locality, strategies to increase Instruction-Level Parallelism (ILP), and NEON-based Single Instruction Multiple Data (SIMD) vectorization to exploit data-level parallelism. Our scalar implementation achieves up to a 5.98x performance increase over a traditional Ternary Compressed Sparse Column (TCSC) baseline for large matrices with 50% ternary nonzero values (sparsity), reaching up to a 50.2% of the processor's theoretical peak performance, and remains stable across varying sparsity levels. Our vectorized implementation delivers up to a 5.59x performance increase for large matrices with 25% sparsity, and remains stable across varying sparsity levels.

Several learned policies have been proposed to replace heuristics for scheduling, caching, and other system components in modern systems. By leveraging diverse features, learning from historical trends, and predicting future behaviors, such models promise to keep pace with ever-increasing workload dynamism and continuous hardware evolution. However, policies trained in isolation may still achieve suboptimal performance when placed together. In this paper, we inspect one such instance in the domain of hardware caching - for the policies of cache replacement and prefetching. We argue that these two policies are bidirection-ally interdependent and make the case for training the two jointly. We propose a joint learning approach based on developing shared representations for the features used by the two policies. We present two approaches to develop these shared representations, one based on a joint encoder and another based on contrastive learning of the embeddings, and demonstrate promising preliminary results for both of these. Finally, we lay down an agenda for future research in this direction.

Memory prefetching has long boosted CPU caches and is increasingly vital for far-memory systems, where large portions of memory are offloaded to cheaper, remote tiers. While effective prefetching requires accurate prediction of future accesses, prior ML approaches have been limited to simulation or small-scale hardware. We introduce FarSight, the first Linux-based far-memory system to leverage deep learning by decoupling application semantics from runtime memory layout. This separation enables offline-trained models to predict access patterns over a compact ordinal vocabulary, which are resolved at runtime through lightweight mappings. Across four data-intensive workloads, FarSight delivers up to 3.6x higher performance than the state-of-the-art.

The global model of differential privacy, which assumes that users trust the data collector, provides strong privacy guarantees and introduces small errors in the output.