English
Related papers

Related papers: Initializing Models with Larger Ones

200 papers

Not all learnable parameters (e.g., weights) contribute equally to a neural network's decision function. In fact, entire layers' parameters can sometimes be reset to random values with little to no impact on the model's decisions. We…

Computer Vision and Pattern Recognition · Computer Science 2024-10-21 Paul Gavrikov , Shashank Agnihotri , Margret Keuper , Janis Keuper

Knowledge embedded in the weights of the artificial neural network can be used to improve the network structure, such as in network compression. However, the knowledge is set up by hand, which may not be very accurate, and relevant…

Neural and Evolutionary Computing · Computer Science 2021-10-13 Mengqiao Han , Xiabi Liu , Zhaoyang Hai , Xin Duan

The aim of this paper is to introduce two widely applicable regularization methods based on the direct modification of weight matrices. The first method, Weight Reinitialization, utilizes a simplified Bayesian assumption with partially…

Machine Learning · Computer Science 2022-06-07 Patrik Reizinger , Bálint Gyires-Tóth

Deep learning models suffer from catastrophic forgetting when trained in an incremental learning setting. In this work, we propose a novel approach to address the task incremental learning problem, which involves training a model on new…

Computer Vision and Pattern Recognition · Computer Science 2021-04-01 Pravendra Singh , Pratik Mazumder , Piyush Rai , Vinay P. Namboodiri

Machine learning requires exuberant amounts of data and computation. Also, models require equally excessive growth in the number of parameters. It is, therefore, sensible to look for technologies that reduce these demands on resources.…

Machine Learning · Computer Science 2023-03-29 Danko Nikolić , Davor Andrić , Vjekoslav Nikolić

Deep Neural Networks are highly over-parameterized and the size of the neural networks can be reduced significantly after training without any decrease in performance. One can clearly see this phenomenon in a wide range of architectures…

Machine Learning · Computer Science 2018-06-19 Utku Evci

Deep neural network (DNN) quantization for fast, efficient inference has been an important tool in limiting the cost of machine learning (ML) model inference. Quantization-specific model development techniques such as regularization,…

Computer Vision and Pattern Recognition · Computer Science 2025-06-13 Stone Yun , Alexander Wong

The interpretability of neural networks (NNs) is a challenging but essential topic for transparency in the decision-making process using machine learning. One of the reasons for the lack of interpretability is random weight initialization,…

Machine Learning · Computer Science 2021-03-01 Shohei Kubota , Hideaki Hayashi , Tomohiro Hayase , Seiichi Uchida

First-break picking is an essential step in seismic data processing. First arrivals should be picked by an expert. This is a time-consuming procedure and subjective to a certain degree, leading to different results for different operators.…

Initializing with pre-trained models when learning on downstream tasks is becoming standard practice in machine learning. Several recent works explore the benefits of pre-trained initialization in a federated learning (FL) setting, where…

Machine Learning · Computer Science 2025-02-13 Divyansh Jhunjhunwala , Pranay Sharma , Zheng Xu , Gauri Joshi

How do we transfer the relevant knowledge from ever larger foundation models into small, task-specific downstream models that can run at much lower costs? Standard transfer learning using pre-trained weights as the initialization transfers…

Machine Learning · Computer Science 2024-06-12 Shikai Qiu , Boran Han , Danielle C. Maddix , Shuai Zhang , Yuyang Wang , Andrew Gordon Wilson

Weight initialization governs signal propagation and gradient flow at the start of training. This paper offers a theory-grounded and empirically validated study across two regimes: compact ReLU multilayer perceptrons and GPT-2-style…

Machine Learning · Computer Science 2025-10-13 Yankun Han

Prior work has demonstrated a consistent tendency in neural networks engaged in continual learning tasks, wherein intermediate task similarity results in the highest levels of catastrophic interference. This phenomenon is attributed to the…

We propose a modularization method that decomposes a deep neural network (DNN) into small modules from a functionality perspective and recomposes them into a new model for some other task. Decomposed modules are expected to have the…

Machine Learning · Computer Science 2021-12-28 Hiroaki Kingetsu , Kenichi Kobayashi , Taiji Suzuki

In this paper, a new method was developed for initialising artificial neural networks predicting dynamics of time series. Initial weighting coefficients were determined for neurons analogously to the case of a linear prediction filter.…

Neural and Evolutionary Computing · Computer Science 2014-10-24 Vladimir V. Bochkarev , Yulia S. Maslennikova

Machine learning is evolving towards high-order models that necessitate pre-training on extensive datasets, a process associated with significant overheads. Traditional models, despite having pre-trained weights, are becoming obsolete due…

Machine Learning · Computer Science 2024-05-10 Chenhui Xu , Xinyao Wang , Fuxun Yu , Jinjun Xiong , Xiang Chen

Pruning on neural networks before training not only compresses the original models, but also accelerates the network training phase, which has substantial application value. The current work focuses on fine-grained pruning, which uses…

Machine Learning · Computer Science 2022-09-28 Xiatao Kang , Ping Li , Jiayi Yao , Chengxi Li

In this paper, orthogonal to the existing data and model studies, we instead resort our efforts to investigate the potential of loss function in a new perspective and present our belief ``Random Weights Networks can Be Acted as Loss Prior…

Computer Vision and Pattern Recognition · Computer Science 2023-03-30 Man Zhou , Naishan Zheng , Jie Huang , Xiangyu Rui , Chunle Guo , Deyu Meng , Chongyi Li , Jinwei Gu

Modern large-scale neural networks are often trained and released in multiple sizes to accommodate diverse inference budgets. To improve efficiency, recent work has explored model upscaling: initializing larger models from trained smaller…

Machine Learning · Computer Science 2026-02-12 Yuxin Ma , Nan Chen , Mateo Díaz , Soufiane Hayou , Dmitriy Kunisky , Soledad Villar

Word embeddings are the interface between the world of discrete units of text processing and the continuous, differentiable world of neural networks. In this work, we examine various random and pretrained initialization methods for…

Computation and Language · Computer Science 2017-11-28 Tom Kocmi , Ondřej Bojar