BeFit: Shifting Complexity from Weights to Bernstein Activations for Resource-Constrained Sensing Researchers introduced BeFit, an end-to-end framework that shifts model capacity from weights to learnable Bernstein polynomial activations for neural networks deployed on resource-constrained microcontrollers. On human activity recognition, Bernstein-activated CNNs (BeCNNs) were 2.75 percentage points more accurate on average than ReLU CNNs of identical architecture, and on a TI CC3220SF (Cortex-M4) weight projection sped BeCNNs up by 2.2x on average while losing 0.8 pp, versus 2.4 pp for ReLU CNNs. Against the MLPerf Tiny DS-CNN reference model on TensorFlow Lite Micro with CMSIS-NN, the unprojected BeCNN achieved 1.0 pp higher accuracy, 2.2x lower latency and 6.4x lower flash, while the projected model ran 5.9x faster and stayed within 1.0 pp of reference accuracy; in RISC-V emulation without a hardware multiplier, instruction-count reduction rose from 5.2x on the Cortex-M4 to 147x. arXiv:2611.23520v1 Announce Type: new Abstract: Time-series sensing increasingly relies on neural networks, yet deployment on resource-constrained microcontrollers remains challenging. Aggressive model compression can reduce accuracy, while restricting weights to multiplication-free representations introduces further losses. We present BeFit, an end-to-end framework that shifts model capacity from weights to learnable Bernstein polynomial activations. BeFit i finds compact Bernstein-activated convolutional neural networks BeCNNs through hardware-aware neural architecture search and trains them via knowledge distillation; ii projects their weights to multiplication-free codes, freezes the codes, and recovers the resulting accuracy loss through the learnable activations; and iii generates model-specific kernels that exploit the projected weights for multiplication-free convolutions. On human activity recognition, BeCNNs are 2.75 percentage points pp more accurate on average than ReLU CNNs of identical architecture. On a TI CC3220SF Cortex-M4 , weight projection speeds BeCNNs up by 2.2x on average while losing 0.8 pp, compared with 2.4 pp for ReLU CNNs. Against the MLPerf Tiny DS-CNN reference model running on TensorFlow Lite Micro with CMSIS-NN, the unprojected BeCNN achieves 1.0 pp higher accuracy, 2.2x lower latency, and 6.4x lower flash, while the projected model remains within 1.0 pp of the reference accuracy and runs 5.9x faster. In emulation on a RISC-V core without a hardware multiplier, the instruction-count reduction relative to the DS-CNN increases from 5.2x on the Cortex-M4 to 147x. BeFit also extends to transformers and MLPs, and across all tasks the final models stay within 1.56 pp of their float32 teachers on average.