Otters++: Energy-Efficient Optical Spiking Transformer Using Natural Device Decay
Researchers developed Otters++, an optical spiking neural network that leverages natural signal decay in optoelectronic devices to implement time-to-first-spike coding more efficiently. The approach converts a hardware characteristic typically considered a limitation into a computational advantage, eliminating the need for explicit digital decay calculations. The work demonstrates that physically grounded spiking transformers can achieve competitive performance on language tasks while maintaining significant energy efficiency gains.
Otters++ addresses a key inefficiency in spiking neural networks (SNNs) by repurposing the natural decay of optoelectronic devices as the primary computation mechanism for time-to-first-spike (TTFS) coding. Rather than computing temporal decay terms digitally, the system uses measured decay characteristics of custom In₂O₃ optoelectronic synapses to directly realize the TTFS temporal component. To scale this approach to Transformer architectures, the researchers established functional equivalence between Otters++ and quantized neural networks, enabling a hybrid training method that uses device-faithful SNN computation in the forward pass and QNN gradients in the backward pass, combined with model distillation. The training process accounts for measured device noise through run-to-run variation sampling and refines energy modeling by considering device sharing and multi-hop communication. On the GLUE benchmark, Otters++ achieved 84.17% average score while maintaining clear energy advantages over prior spiking Transformer baselines.
What's missing
The paper does not provide detailed comparisons of absolute energy consumption (in joules or watts) against standard transformer baselines, instead focusing on relative advantages over prior spiking approaches. Additionally, the scalability of the In₂O₃ optoelectronic synapse fabrication process and its practical deployment constraints are not discussed. The study also does not address latency implications of the TTFS approach compared to standard transformers.
What different sources said
- arXiv cs.AICenter
Otters++: A Time-to-first-spike Based Energy Efficient Optical Spiking Transformer
Related
Topology-Aware Thermodynamics Improves DNA Probe Specificity Design
Researchers developed a new framework for designing DNA probes that accounts for the spatial organization of matched sequences, not just overall thermodynamic stability. Traditional methods rely on scalar measures like melting temperature and free energy, which miss how mismatches are distributed along the probe. The approach could improve diagnostic accuracy in applications like HPV detection and gene expression profiling.
Study Identifies Optimal Thermal Dose for Combining Focused Ultrasound with Immunotherapy in Tumors
Researchers used multimodal PET imaging to identify an optimal thermal dose range for focused ultrasound ablation that destroys tumor tissue while preserving conditions for immunotherapy delivery. The study found that excessive heating collapses blood vessels needed for antibody access, while insufficient heating fails to adequately reduce tumor burden. The findings could guide clinical design of combination treatments pairing thermal ablation with immunotherapies.
Plant MSH1 Protein Functions as Mismatch-Directed Nuclease for Organelle Genome Maintenance
Researchers have identified the precise mechanism by which the AtMSH1 protein in Arabidopsis plants recognizes and cleaves DNA mismatches and lesions, preventing mutations in organellar genomes. The protein combines a DNA mismatch recognition module with a nuclease domain that makes staggered cuts at specific positions relative to DNA damage. This discovery explains how plants maintain unusually low mutation rates in their mitochondrial and chloroplast DNA compared to other eukaryotes.