Elastoformer: Enabling Dynamic Adaptivity via Elastic Model Transformation

• URL: http://arxiv.org/abs/2609.10018v1
• Date: Wed, 09 Sep 2026 10:52:21 GMT
• Title: Elastoformer: Enabling Dynamic Adaptivity via Elastic Model Transformation
• Abstract summary: Elastoformer is a framework that transforms conventional neural networks (NN) into Elastic NN capable of real-time elastic inference.<n>Our framework achieves up to 85% reduction in FLOPs, 50% reduction in latency and 76% reduction in memory overhead.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: EdgeAI systems are increasingly employing computer vision applications to enable intelligent, on-device decision-making in real-time. However, these deployments face highly dynamic operational conditions, with fluctuating constraints on latency, power availability, and memory resources. Deep Neural Networks (DNN), which follow fixed computational execution flows, lack the flexibility to adapt to such variability, resulting in inefficient and suboptimal performance in edge scenarios. This underscores the need for architectures that are not only efficient but also dynamically scalable at runtime. In this paper, we propose Elastoformer: A framework that transforms conventional neural networks (NN) into Elastic NN capable of real-time elastic inference. Unlike the conventional bag-of-models approach, which requires maintaining multiple independent models for different operating conditions, Elastoformer offers a single, modular solution that dynamically switches between multiple modes of operation at runtime, adapting efficiently to the changing computational budgets of edge devices without the overhead of managing separate models. Experiments reveal that our framework achieves up to 85% reduction in computation FLOPs, 50% reduction in latency and 76% reduction in memory overhead, while showcasing the architecture agnostic nature of the framework across both Vision Transformers and CNNs. Our code is available at https://github.com/sudaksh14/Elastoformer.

Related papers

• ElasticDiT: Efficient Diffusion Transformers via Elastic Architecture and Sparse Attention for High-Resolution Image Generation on Mobile Devices [19.789749822094617]
  Diffusion Transformer (DiT) architecture is the state-of-the-art paradigm for high-fidelity image generation.<n>However, deploying these models on resource-constrained mobile devices entails prohibitive computational and memory overhead.<n>We introduce ElasticDiT, which achieves this dynamic trade-off by adjusting spatial compression ratios and DiT block depths.
  arXiv  Detail & Related papers  (2026-05-15T07:13:27Z)
• Elastic Spiking Transformers for Efficient Gesture Understanding [6.52225846892392]
  Spiking Neural Networks (SNNs) offer energy-efficient processing of event-based sensor data for healthcare applications.<n>We introduce the Elastic Spiking Transformer, a runtime-adaptive architecture that brings elasticity into the spiking paradigm.<n>We evaluate the approach on CIFAR10/100, CIFAR10-DVS, and the EHWGesture clinical gesture understanding dataset.
  arXiv  Detail & Related papers  (2026-05-04T12:35:52Z)
• Switchable Activation Networks [25.854474675239842]
  SWAN (Switchable Activation Networks) is a framework that equips each neural unit with a deterministic, input-dependent binary gate.<n>This dynamic control mechanism allocates computation adaptively, reducing redundancy while preserving accuracy.<n>By reframing efficiency as a problem of learned activation control, SWAN unifies the strengths of sparsity, pruning, and adaptive inference.
  arXiv  Detail & Related papers  (2026-02-17T08:14:10Z)
• Adaptive Mesh-Quantization for Neural PDE Solvers [51.26961483962011]
  Graph Neural Networks can handle the irregular meshes required for complex geometries and boundary conditions, but still apply uniform computational effort across all nodes.<n>We propose Adaptive Mesh Quantization: spatially adaptive quantization across mesh node, edge, and cluster features, dynamically adjusting the bit-width used by a quantized model.<n>We demonstrate our framework's effectiveness by integrating it with two state-of-the-art models, MP-PDE and GraphViT, to evaluate performance across multiple tasks.
  arXiv  Detail & Related papers  (2025-11-23T14:47:24Z)
• Elastic ViTs from Pretrained Models without Retraining [74.5386166956142]
  Vision foundation models achieve remarkable performance but are only available in a limited set of pre-determined sizes.<n>We introduce SnapViT: Single-shot network approximation for pruned Vision Transformers.<n>Our approach efficiently combines gradient information with cross-network structure correlations, approximated via an evolutionary algorithm.
  arXiv  Detail & Related papers  (2025-10-20T16:15:03Z)
• Deep Hierarchical Learning with Nested Subspace Networks [53.71337604556311]
  We propose Nested Subspace Networks (NSNs) for large neural networks.<n>NSNs enable a single model to be dynamically and granularly adjusted across a continuous spectrum of compute budgets.<n>We show that NSNs can be surgically applied to pre-trained LLMs and unlock a smooth and predictable compute-performance frontier.
  arXiv  Detail & Related papers  (2025-09-22T15:13:14Z)
• ReStNet: A Reusable & Stitchable Network for Dynamic Adaptation on IoT Devices [16.762206782460296]
  ReStNet dynamically constructs a hybrid network by stitching two pre-trained models together.<n>It achieves flexible accuracy-efficiency trade-offs at runtime while significantly reducing training cost.
  arXiv  Detail & Related papers  (2025-06-08T16:14:37Z)
• Towards Adaptive Deep Learning: Model Elasticity via Prune-and-Grow CNN Architectures [0.0]
  This thesis explores methods for enabling CNNs to dynamically adjust their computational complexity based on available hardware resources.<n>We introduce adaptive CNN architectures capable of scaling their capacity at runtime, thus efficiently balancing performance and resource utilization.
  arXiv  Detail & Related papers  (2025-05-16T10:06:55Z)
• DyCE: Dynamically Configurable Exiting for Deep Learning Compression and Real-time Scaling [1.8350044465969415]
  DyCE can adjust the performance-complexity trade-off of a deep learning model at runtime without requiring re-initialization or redeployment on inference hardware.<n>DyCE significantly reduces computational complexity by 23.5% for ResNet152 and 25.9% for ConvNextv2-tiny on ImageNet, with accuracy reductions of less than 0.5%.
  arXiv  Detail & Related papers  (2024-03-04T03:09:28Z)
• Efficient and Flexible Neural Network Training through Layer-wise Feedback Propagation [49.44309457870649]
  Layer-wise Feedback feedback (LFP) is a novel training principle for neural network-like predictors.<n>LFP decomposes a reward to individual neurons based on their respective contributions.<n>Our method then implements a greedy reinforcing approach helpful parts of the network and weakening harmful ones.
  arXiv  Detail & Related papers  (2023-08-23T10:48:28Z)
• Energy-efficient Task Adaptation for NLP Edge Inference Leveraging Heterogeneous Memory Architectures [68.91874045918112]
  adapter-ALBERT is an efficient model optimization for maximal data reuse across different tasks. We demonstrate the advantage of mapping the model to a heterogeneous on-chip memory architecture by performing simulations on a validated NLP edge accelerator.
  arXiv  Detail & Related papers  (2023-03-25T14:40:59Z)
• Slimmable Domain Adaptation [112.19652651687402]
  We introduce a simple framework, Slimmable Domain Adaptation, to improve cross-domain generalization with a weight-sharing model bank. Our framework surpasses other competing approaches by a very large margin on multiple benchmarks.
  arXiv  Detail & Related papers  (2022-06-14T06:28:04Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.