Search arXivSearch

arXiv · 2512.04705

Hardware-Algorithm Co-Optimization of Early-Exit Neural Networks for Multi-Core Edge Accelerators

Abstract

Deployment of dynamic neural networks on edge accelerators requires careful consideration of hardware constraints beyond conventional complexity metrics such as Multiply-Accumulate operations. In Early-Exiting Neural Networks (EENN), exit placement, quantization level, and hardware workload mapping interact in non-trivial ways, influencing memory traffic, accelerator utilization, and ultimately energy-latency trade-offs. These interactions remain insufficiently understood in existing Neural Architecture Search (NAS) approaches, which typically rely on proxy metrics or hardware-in-the-loop evaluation. This work presents a hardware-algorithm co-design framework for EENN that explicitly models the interplay between quantization, exit configuration, and multi-core accelerator mapping. Using analytical design space exploration, we characterize how small architectural variations can induce disproportionate changes in hardware efficiency due to tensor dimension alignment and dataflow effects. Building on this analysis, we formulate EENN deployment as a constrained multi-objective optimization problem balancing accuracy, energy-latency product, exit overhead, and dynamic inference behavior. Experimental results on CIFAR-10 demonstrate that the proposed framework identifies architectures achieving over 50\% reduction in energy-latency product compared to static baselines under 8-bit quantization. The results highlight the importance of deployment-aware co-design for dynamic inference on heterogeneous edge platforms.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Alaa Zniber, Arne Symons, Ouassim Karrakchou, Marian Verhelst, Mounir Ghogho. 2026-03-31. Hardware-Algorithm Co-Optimization of Early-Exit Neural Networks for Multi-Core Edge Accelerators. https://arxiv.org/abs/2512.04705

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

Counting Small Induced Subgraphs: Hardness of Symmetry-Based Properties

Jerrum and Meeks (TOCT, JCSS 2015) introduced the counting problems $\text{IndSub}(Φ)$ for fixed graph properties $Φ$: Given an input graph $G$ and $k\in\mathbb N$, count the $k$-vertex subsets $S \subseteq V(G)$ such that the induced subgraph $G[S]$ satisfies $Φ$. For recursively enumerable $Φ$, it is known that $\text{IndSub}(Φ)$ is either #W[1]-hard or fixed-parameter tractable. A direct classification depending on $Φ$ however still remains open. In particular, the status was open for the property of graphs without nontrivial automorphisms, also mentioned in a very recent survey on parameterized counting by Roth (Comput.~Sci.~Rev.~2026). This is a natural property that evades all currently known techniques for proving #W[1]-hardness, including a general toolkit based on Fourier analysis that was very recently introduced by Curticapean and Neuen (SODA~2025). In this paper, we show that counting induced $k$-vertex graphs without nontrivial automorphisms is #W[1]-hard by constructing ``clique scaffolds'', i.e., problem-specific restrictions of the property that enable a reduction from the $k$-clique problem. More generally, we show that for every finite group $Q$, counting $k$-vertex induced subgraphs with automorphism group $Q$ is #W[1]-hard.

cs.CC

Improved Algorithms for the Remote Point Problem

The Remote Point Problem (RPP) is an algorithmic problem that asks, given a linear subspace $L \subseteq \mathbb{F}^n$ of dimension $k$, to deterministically find a vector $v \in \mathbb{F}^n$ far in Hamming distance from $L$. This problem was introduced by Alon, Panigrahy and Yekhanin [APY09], motivated in part by the matrix rigidity approach for proving circuit lower bounds. An algorithm is said to achieve remoteness $d$ if it finds a vector $v$ whose Hamming distance from $L$ is at least $d$. We observe that over the rational numbers, the problem admits a deterministic polynomial-time algorithm that achieves optimal remoteness $n-k$. Over finite fields, we obtain a (modest) improvement of a result of Alon, Panigrahy and Yekhanin [APY09], and give an algorithm that achieves remoteness $Ω\left(\frac{n}{\max\{k, \log n\}} \log n\right)$.

cs.CC

Strong Selective and List-Decoding Direct Product Theorems for Quantum Query Complexity

Quantum strong direct-product theorems for specific functions have been known for nearly two decades. These have been extended to general results for function computation and state generation. The proofs of these results use a version of the multiplicative adversary method that does not naturally extend to relations. Standard strong direct-product theorems apply when algorithms must correctly answer every given question. Prior work extended them to equivalent threshold direct-product theorems, which require answers to all questions but only require that most answers are correct. We focus on two further generalizations. Strong selective direct-products apply to algorithms that adaptively choose, based on what they learn from queries, which questions from a large list to answer. This generalization is relational and useful for proving time-space tradeoffs. We prove a quantum strong selective direct-product theorem for all functions using a new multiplicative adversary formulation for relations that satisfies a strong selective direct product property while being strong enough to capture any query lower bound for functions proven by negative-weights adversaries. This was not previously known even without selectivity. The second generalization is list-decoding direct product problems introduced by Ben-David and Blais for classical randomized query complexity. These allow an algorithm to produce a large list of possible output vectors such that one of them is fully correct. They proved that such theorems hold for classical randomized complexity of all Boolean functions. We prove a quantum analogue of this theorem for all partial Boolean functions. We show that strong list-decoding direct-product theorems are implied by a special case of multiplicative adversaries which we show, via a new reduction, can be obtained from negative-weights adversaries for any Boolean-valued function.

cs.CC