Large-batch Contrastive Learning (CL), the foundation of modern representation learning, is fundamentally incompatible with the volatile resource constraints of edge devices. This conflict creates a dilemma: small on-device batches degrade model fidelity, while offloading to the cloud incurs unacceptable latency and bandwidth costs. Existing solutions often resort to static model compression, which fails to adapt to the runtime volatility of edge environments. To bridge this gap, we present StreamSplit, a novel framework that makes streaming CL practical across heterogeneous ARM client platforms. StreamSplit resolves the conflict between the continuous nature of ambient audio and the discrete batch requirements of models like CLAP and COLA. We introduce: (1) A distribution-based streaming framework that decouples representation quality from local batch size, using a tractable Hybrid Loss to maintain fidelity despite sparse updates; and (2) An Uncertainty-Guided Adaptive Splitter that uses a lightweight Reinforcement Learning (RL) policy to dynamically partition computation. Uniquely, this policy integrates real-time resource monitoring with embedding ambiguity to optimize the accuracy-latency trade-off on the fly. We evaluate StreamSplit on diverse hardware, from the resource-constrained Raspberry Pi 4 to the high-performance Apple M2. Results demonstrate that StreamSplit reduces per-sample latency by up to 4.7× and cuts bandwidth by 77.1% and energy by 52.3% compared to server-centric baselines. Crucially, it maintains accuracy within 2.2% of server-centric models, proving that adaptive, distributed learning is a viable path for the modern edge ecosystem.

Diffuse waste data and associated privacy concerns present significant challenges for effective waste classification in the Internet of Things (IoT) realm. This research introduces a novel approach that leverages differential privacy (DP) and federated transfer learning (FTL) to address the issues, enabling waste classification while preserving privacy within the IoT ecosystem. By integrating Federated Learning (FL), Transfer Learning (TL), and DP, our proposed method facilitates collaborative training while ensuring data privacy. In this methodology, a pre-trained model, initially trained on the ImageNet dataset, is disseminated to IoT devices. Subsequently, these devices perform local training using the TrashNet and Garbage Classification datasets. This process allows devices to capture waste characteristics unique to their individual environments. Through the fusion of general knowledge pertaining to the trained model and local insights, the proposed approach achieves efficient waste classification. The study critically examines the implications for privacy, biases resulting from limited local data, and trade-offs between privacy and model performance. The experimental evaluation demonstrates the effectiveness of the approach and underscores the importance of ensuring privacy-sensitive waste classification. This research contributes to the discourse on FTL and encourages further research into privacy-preserving waste classification within the IoT.