Across various domains, the growing advocacy for open science and open-source machine learning has made an increasing number of models publicly available. These models allow practitioners to integrate them into their own contexts, reducing the need for extensive data labeling, training, and calibration. However, selecting the best model for a specific target population remains challenging due to issues like limited transferability, data heterogeneity, and the difficulty of obtaining true labels or outcomes in real-world settings. In this paper, we propose an unsupervised model aggregation method, U-aggregation, designed to integrate multiple pre-trained models for enhanced and robust performance in new populations. Unlike existing supervised model aggregation or super learner approaches, U-aggregation assumes no observed labels or outcomes in the target population. Our method addresses limitations in existing unsupervised model aggregation techniques by accommodating more realistic settings, including heteroskedasticity at both the model and individual levels, and the presence of adversarial models. Drawing on insights from random matrix theory, U-aggregation incorporates a variance stabilization step and an iterative sparse signal recovery process. These steps improve the estimation of individuals' true underlying risks in the target population and evaluate the relative performance of candidate models. We provide a theoretical investigation and systematic numerical experiments to elucidate the properties of U-aggregation. We demonstrate its potential real-world application by using U-aggregation to enhance genetic risk prediction of complex traits, leveraging publicly available models from the PGS Catalog.

A common task arising in various domains is that of ranking items based on the outcomes of pairwise comparisons, from ranking players and teams in sports to ranking products or brands in marketing studies and recommendation systems. Statistical inference-based methods such as the Bradley-Terry model, which extract rankings based on an underlying generative model of the comparison outcomes, have emerged as flexible and powerful tools to tackle the task of ranking in empirical data. In situations with limited and/or noisy comparisons, it is often challenging to confidently distinguish the performance of different items based on the evidence available in the data. However, existing inference-based ranking methods overwhelmingly choose to assign each item to a unique rank or score, suggesting a meaningful distinction when there is none. Here, we address this problem by developing a principled Bayesian methodology for learning partial rankings -- rankings with ties -- that distinguishes among the ranks of different items only when there is sufficient evidence available in the data. Our framework is adaptable to any statistical ranking method in which the outcomes of pairwise observations depend on the ranks or scores of the items being compared. We develop a fast agglomerative algorithm to perform Maximum A Posteriori (MAP) inference of partial rankings under our framework and examine the performance of our method on a variety of real and synthetic network datasets, finding that it frequently gives a more parsimonious summary of the data than traditional ranking, particularly when observations are sparse.

The growing concern about data privacy has led to the development of private inference (PI) frameworks in client-server applications which protects both data privacy and model IP. However, the cryptographic primitives required yield significant latency overhead which limits its wide-spread application. At the same time, changing environments demand the PI service to be robust against various naturally occurring and gradient-based perturbations. Despite several works focused on the development of latency-efficient models suitable for PI, the impact of these models on robustness has remained unexplored. Towards this goal, this paper presents RLNet, a class of robust linearized networks that can yield latency improvement via reduction of high-latency ReLU operations while improving the model performance on both clean and corrupted images. In particular, RLNet models provide a "triple win ticket" of improved classification accuracy on clean, naturally perturbed, and gradient-based perturbed images using a shared-mask shared-weight architecture with over an order of magnitude fewer ReLUs than baseline models. To demonstrate the efficacy of RLNet, we perform extensive experiments with ResNet and WRN model variants on CIFAR-10, CIFAR-100, and Tiny-ImageNet datasets. Our experimental evaluations show that RLNet can yield models with up to 11.14x fewer ReLUs, with accuracy close to the all-ReLU models, on clean, naturally perturbed, and gradient-based perturbed images. Compared with the SoTA non-robust linearized models at similar ReLU budgets, RLNet achieves an improvement in adversarial accuracy of up to ~47%, naturally perturbed accuracy up to ~16.4%, while improving clean image accuracy up to ~1.5%.

Punctuation restoration is an important task in automatic speech recognition (ASR) which aim to restore the syntactic structure of generated ASR texts to improve readability. While punctuated texts are abundant from written documents, the discrepancy between written punctuated texts and ASR texts limits the usability of written texts in training punctuation restoration systems for ASR texts. This paper proposes a reinforcement learning method to exploit in-topic written texts and recent advances in large pre-trained generative language models to bridge this gap. The experiments show that our method achieves state-of-the-art performance on the ASR test set on two benchmark datasets for punctuation restoration.

Large number of ReLU and MAC operations of Deep neural networks make them ill-suited for latency and compute-efficient private inference. In this paper, we present a model optimization method that allows a model to learn to be shallow. In particular, we leverage the ReLU sensitivity of a convolutional block to remove a ReLU layer and merge its succeeding and preceding convolution layers to a shallow block. Unlike existing ReLU reduction methods, our joint reduction method can yield models with improved reduction of both ReLUs and linear operations by up to 1.73x and 1.47x, respectively, evaluated with ResNet18 on CIFAR-100 without any significant accuracy-drop.

Deep Learning (DL) has transformed the automation of a wide range of industries and finds increasing ubiquity in society. The high complexity of DL models and its widespread adoption has led to global energy consumption doubling every 3-4 months. Currently, the relationship between the DL model configuration and energy consumption is not well established. At a general computational energy model level, there is both strong dependency to both the hardware architecture (e.g. generic processors with different configuration of inner components- CPU and GPU, programmable integrated circuits - FPGA), as well as different interacting energy consumption aspects (e.g., data movement, calculation, control). At the DL model level, we need to translate non-linear activation functions and its interaction with data into calculation tasks. Current methods mainly linearize nonlinear DL models to approximate its theoretical FLOPs and MACs as a proxy for energy consumption. Yet, this is inaccurate (est. 93\% accuracy) due to the highly nonlinear nature of many convolutional neural networks (CNNs) for example. In this paper, we develop a bottom-level Transistor Operations (TOs) method to expose the role of non-linear activation functions and neural network structure in energy consumption. We translate a range of feedforward and CNN models into ALU calculation tasks and then TO steps. This is then statistically linked to real energy consumption values via a regression model for different hardware configurations and data sets. We show that our proposed TOs method can achieve a 93.61% - 99.51% precision in predicting its energy consumption.

With large quantities of data typically available nowadays, forecasting models that are trained across sets of time series, known as Global Forecasting Models (GFM), are regularly outperforming traditional univariate forecasting models that work on isolated series. As GFMs usually share the same set of parameters across all time series, they often have the problem of not being localised enough to a particular series, especially in situations where datasets are heterogeneous. We study how ensembling techniques can be used with generic GFMs and univariate models to solve this issue. Our work systematises and compares relevant current approaches, namely clustering series and training separate submodels per cluster, the so-called ensemble of specialists approach, and building heterogeneous ensembles of global and local models. We fill some gaps in the approaches and generalise them to different underlying GFM model types. We then propose a new methodology of clustered ensembles where we train multiple GFMs on different clusters of series, obtained by changing the number of clusters and cluster seeds. Using Feed-forward Neural Networks, Recurrent Neural Networks, and Pooled Regression models as the underlying GFMs, in our evaluation on six publicly available datasets, the proposed models are able to achieve significantly higher accuracy than baseline GFM models and univariate forecasting methods.

In the current context of Big Data, the nature of many forecasting problems has changed from predicting isolated time series to predicting many time series from similar sources. This has opened up the opportunity to develop competitive global forecasting models that simultaneously learn from many time series. But, it still remains unclear when global forecasting models can outperform the univariate benchmarks, especially along the dimensions of the homogeneity/heterogeneity of series, the complexity of patterns in the series, the complexity of forecasting models, and the lengths/number of series. Our study attempts to address this problem through investigating the effect from these factors, by simulating a number of datasets that have controllable time series characteristics. Specifically, we simulate time series from simple data generating processes (DGP), such as Auto Regressive (AR) and Seasonal AR, to complex DGPs, such as Chaotic Logistic Map, Self-Exciting Threshold Auto-Regressive, and Mackey-Glass Equations. The data heterogeneity is introduced by mixing time series generated from several DGPs into a single dataset. The lengths and the number of series in the dataset are varied in different scenarios. We perform experiments on these datasets using global forecasting models including Recurrent Neural Networks (RNN), Feed-Forward Neural Networks, Pooled Regression (PR) models and Light Gradient Boosting Models (LGBM), and compare their performance against standard statistical univariate forecasting techniques. Our experiments demonstrate that when trained as global forecasting models, techniques such as RNNs and LGBMs, which have complex non-linear modelling capabilities, are competitive methods in general under challenging forecasting scenarios such as series having short lengths, datasets with heterogeneous series and having minimal prior knowledge of the patterns of the series.