SLoRA: Federated Parameter Efficient Fine-Tuning of Language Models

Transfer learning via fine-tuning pre-trained transformer models has gained significant success in delivering state-of-the-art results across various NLP tasks. In the absence of centralized data, Federated Learning (FL) can benefit from distributed and private data of the FL edge clients for fine-tuning. However, due to the limited communication, computation, and storage capabilities of edge devices and the huge sizes of popular transformer models, efficient fine-tuning is crucial to make federated training feasible. This work explores the opportunities and challenges associated with applying parameter efficient fine-tuning (PEFT) methods in different FL settings for language tasks. Specifically, our investigation reveals that as the data across users becomes more diverse, the gap between fully fine-tuning the model and employing PEFT methods widens. To bridge this performance gap, we propose a method called SLoRA, which overcomes the key limitations of LoRA in high heterogeneous data scenarios through a novel data-driven initialization technique. Our experimental results demonstrate that SLoRA achieves performance comparable to full fine-tuning, with significant sparse updates with approximately $\sim 1\%$ density while reducing training time by up to $90\%$.

Machine Collaboration

We propose a new ensemble framework for supervised learning, named machine collaboration (MaC), based on a collection of base machines for prediction tasks. Different from bagging/stacking (a parallel & independent framework) and boosting (a sequential & top-down framework), MaC is a type of circular & interactive learning framework. The circular & interactive feature helps the base machines to transfer information circularly and update their own structures and parameters accordingly. The theoretical result on the risk bound of the estimator based on MaC shows that circular & interactive feature can help MaC reduce the risk via a parsimonious ensemble. We conduct extensive experiments on simulated data and 119 benchmark real data sets. The results of the experiments show that in most cases, MaC performs much better than several state-of-the-art methods, including CART, neural network, stacking, and boosting.

Nested Model Averaging on Solution Path for High-dimensional Linear Regression

We study the nested model averaging method on the solution path for a high-dimensional linear regression problem. In particular, we propose to combine model averaging with regularized estimators (e.g., lasso and SLOPE) on the solution path for high-dimensional linear regression. In simulation studies, we first conduct a systematic investigation on the impact of predictor ordering on the behavior of nested model averaging, then show that nested model averaging with lasso and SLOPE compares favorably with other competing methods, including the infeasible lasso and SLOPE with the tuning parameter optimally selected. A real data analysis on predicting the per capita violent crime in the United States shows an outstanding performance of the nested model averaging with lasso.

GSANet: Semantic Segmentation with Global and Selective Attention

This paper proposes a novel deep learning architecture for semantic segmentation. The proposed Global and Selective Attention Network (GSANet) features Atrous Spatial Pyramid Pooling (ASPP) with a novel sparsemax global attention and a novel selective attention that deploys a condensation and diffusion mechanism to aggregate the multi-scale contextual information from the extracted deep features. A selective attention decoder is also proposed to process the GSA-ASPP outputs for optimizing the softmax volume. We are the first to benchmark the performance of semantic segmentation networks with the low-complexity feature extraction network (FXN) MobileNetEdge, that is optimized for low latency on edge devices. We show that GSANet can result in more accurate segmentation with MobileNetEdge, as well as with strong FXNs, such as Xception. GSANet improves the state-of-art semantic segmentation accuracy on both the ADE20k and the Cityscapes datasets.

Some Chances and Challenges in Applying Language Technologies to Historical Studies in Chinese

We report applications of language technology to analyzing historical documents in the Database for the Study of Modern Chinese Thoughts and Literature (DSMCTL). We studied two historical issues with the reported techniques: the conceptualization of "huaren" (Chinese people) and the attempt to institute constitutional monarchy in the late Qing dynasty. We also discuss research challenges for supporting sophisticated issues using our experience with DSMCTL, the Database of Government Officials of the Republic of China, and the Dream of the Red Chamber. Advanced techniques and tools for lexical, syntactic, semantic, and pragmatic processing of language information, along with more thorough data collection, are needed to strengthen the collaboration between historians and computer scientists.