The phenomenon of double descent has challenged the traditional bias-variance trade-off in supervised learning but remains unexplored in unsupervised learning, with some studies arguing for its absence. In this study, we first demonstrate analytically that double descent does not occur in linear unsupervised autoencoders (AEs). In contrast, we show for the first time that both double and triple descent can be observed with nonlinear AEs across various data models and architectural designs. We examine the effects of partial sample and feature noise and highlight the importance of bottleneck size in influencing the double descent curve. Through extensive experiments on both synthetic and real datasets, we uncover model-wise, epoch-wise, and sample-wise double descent across several data types and architectures. Our findings indicate that over-parameterized models not only improve reconstruction but also enhance performance in downstream tasks such as anomaly detection and domain adaptation, highlighting their practical value in complex real-world scenarios.

We study robust parameter-efficient fine-tuning (PEFT) techniques designed to improve accuracy and generalization while operating within strict computational and memory hardware constraints, specifically focusing on large-language models (LLMs). Existing PEFT methods often lack robustness and fail to generalize effectively across diverse tasks, leading to suboptimal performance in real-world scenarios. To address this, we present a new highly computationally efficient framework called AdaZo-SAM, combining Adam and Sharpness-Aware Minimization (SAM) while requiring only a single-gradient computation in every iteration. This is achieved using a stochastic zeroth-order estimation to find SAM's ascent perturbation. We provide a convergence guarantee for AdaZo-SAM and show that it improves the generalization ability of state-of-the-art PEFT methods. Additionally, we design a low-rank gradient optimization method named LORENZA, which is a memory-efficient version of AdaZo-SAM. LORENZA utilizes a randomized SVD scheme to efficiently compute the subspace projection matrix and apply optimization steps onto the selected subspace. This technique enables full-parameter fine-tuning with adaptive low-rank gradient updates, achieving the same reduced memory consumption as gradient-low-rank-projection methods. We provide a convergence analysis of LORENZA and demonstrate its merits for pre-training and fine-tuning LLMs.

The extensive need for computational resources poses a significant obstacle to deploying large-scale Deep Neural Networks (DNN) on devices with constrained resources. At the same time, studies have demonstrated that a significant number of these DNN parameters are redundant and extraneous. In this paper, we introduce a novel approach for learning structured sparse neural networks, aimed at bridging the DNN hardware deployment challenges. We develop a novel regularization technique, termed Weighted Group Sparse Envelope Function (WGSEF), generalizing the Sparse Envelop Function (SEF), to select (or nullify) neuron groups, thereby reducing redundancy and enhancing computational efficiency. The method speeds up inference time and aims to reduce memory demand and power consumption, thanks to its adaptability which lets any hardware specify group definitions, such as filters, channels, filter shapes, layer depths, a single parameter (unstructured), etc. The properties of the WGSEF enable the pre-definition of a desired sparsity level to be achieved at the training convergence. In the case of redundant parameters, this approach maintains negligible network accuracy degradation or can even lead to improvements in accuracy. Our method efficiently computes the WGSEF regularizer and its proximal operator, in a worst-case linear complexity relative to the number of group variables. Employing a proximal-gradient-based optimization technique, to train the model, it tackles the non-convex minimization problem incorporating the neural network loss and the WGSEF. Finally, we experiment and illustrate the efficiency of our proposed method in terms of the compression ratio, accuracy, and inference latency.

Large Language Models (LLMs), with billions of parameters, present significant challenges for full finetuning due to the high computational demands, memory requirements, and impracticality of many real-world applications. When faced with limited computational resources or small datasets, updating all model parameters can often result in overfitting. To address this, lightweight finetuning techniques have been proposed, like learning low-rank adapter layers. These methods aim to train only a few additional parameters combined with the base model, which remains frozen, reducing resource usage and mitigating overfitting risks. In this work, we propose an adaptor model based on stochastic gates that simultaneously sparsify the frozen base model with task-specific adaptation. Our method comes with a small number of trainable parameters and allows us to speed up the base model inference with competitive accuracy. We evaluate it in additional variants by equipping it with additional low-rank parameters and comparing it to several recent baselines. Our results show that the proposed method improves the finetuned model accuracy comparatively to the several baselines and allows the removal of up to 20-40\% without significant accuracy loss.

We propose an efficient method to learn both unstructured and structured sparse neural networks during training, utilizing a novel generalization of the sparse envelope function (SEF) used as a regularizer, termed {\itshape{weighted group sparse envelope function}} (WGSEF). The WGSEF acts as a neuron group selector, which is leveraged to induce structured sparsity. The method ensures a hardware-friendly structured sparsity of a deep neural network (DNN) to efficiently accelerate the DNN's evaluation. Notably, the method is adaptable, letting any hardware specify group definitions, such as filters, channels, filter shapes, layer depths, a single parameter (unstructured), etc. Owing to the WGSEF's properties, the proposed method allows to a pre-define sparsity level that would be achieved at the training convergence, while maintaining negligible network accuracy degradation or even improvement in the case of redundant parameters. We introduce an efficient technique to calculate the exact value of the WGSEF along with its proximal operator in a worst-case complexity of $O(n)$, where $n$ is the total number of group variables. In addition, we propose a proximal-gradient-based optimization method to train the model, that is, the non-convex minimization of the sum of the neural network loss and the WGSEF. Finally, we conduct an experiment and illustrate the efficiency of our proposed technique in terms of the completion ratio, accuracy, and inference latency.

Large Language Models (LLMs) have shown promise in highly-specialized domains, however challenges are still present in aspects of accuracy and costs. These limitations restrict the usage of existing models in domain-specific tasks. While fine-tuning pre-trained models have shown promising results, this process can be computationally expensive and require massive datasets of the specialized application in hand. In this work, we bridge that gap. We have developed Phi-2-Legal and Mistral-Legal-7B, which are language models specifically designed for legal applications. These models are based on Phi-2 and Mistral-7B-v0.1, and have gone through continued pre-training with over 500 million tokens of legal texts. Our innovative approach significantly improves capabilities in legal tasks by using Large Language Models (LLMs) to convert raw training data into reading comprehension text. Our legal LLMs have demonstrated superior performance in legal benchmarks, even outperforming models trained on much larger datasets with more resources. This work emphasizes the effectiveness of continued pre-training on domain-specific texts, while using affordable LLMs for data conversion, which gives these models domain expertise while retaining general language understanding capabilities. While this work uses the legal domain as a test case, our method can be scaled and applied to any pre-training dataset, resulting in significant improvements across different tasks. These findings underscore the potential of domain-adaptive pre-training and reading comprehension for the development of highly effective domain-specific language models.

Social media platforms (SMPs) leverage algorithmic filtering (AF) as a means of selecting the content that constitutes a user's feed with the aim of maximizing their rewards. Selectively choosing the contents to be shown on the user's feed may yield a certain extent of influence, either minor or major, on the user's decision-making, compared to what it would have been under a natural/fair content selection. As we have witnessed over the past decade, algorithmic filtering can cause detrimental side effects, ranging from biasing individual decisions to shaping those of society as a whole, for example, diverting users' attention from whether to get the COVID-19 vaccine or inducing the public to choose a presidential candidate. The government's constant attempts to regulate the adverse effects of AF are often complicated, due to bureaucracy, legal affairs, and financial considerations. On the other hand SMPs seek to monitor their own algorithmic activities to avoid being fined for exceeding the allowable threshold. In this paper, we mathematically formalize this framework and utilize it to construct a data-driven statistical auditing procedure to regulate AF from deflecting users' beliefs over time, along with sample complexity guarantees. This state-of-the-art algorithm can be used either by authorities acting as external regulators or by SMPs for self-auditing.

The paper considers the sparse envelope function, defined as the biconjugate of the sum of a squared ℓ 2 -norm function and the indicator of the set of k -sparse vectors. It is shown that both function and proximal values of the sparse envelope function can be reduced into a one-dimensional search that can be efficiently performed in linear time complexity in expectation. The sparse envelope function naturally serves as a regularizer that can handle both sparsity and grouping information in inverse problems, and can also be utilized in sparse support vector machine problems.

Fault Management (FM) is a cardinal feature in communication networks. One of the most common FM approaches is to use periodic keepalive messages. Hence, switches and routers are required to transmit a large number of FM messages periodically, requiring a hardware-based packet generator that periodically transmits a set of messages that are stored in an expensive on-chip memory. With the rapid growth of carrier networks, and as 5G technologies emerge, the number of users and the traffic rates are expected to significantly increase over the next few years. Consequently, we expect the on-chip memories used for FM to become a costly component in switch and router chips. We introduce a novel approach in which FM messages are stored in compressed form in the on-chip memory, allowing to significantly reduce the memory size. We present FM-Delta, a simple hardware-friendly delta encoding algorithm that allows FM messages to be compressed by a factor of 2.6. We show that this compression ratio is very close to the results of the zlib compression library, which requires much higher implementation complexity.