paper-with-me

Papers

What do larger image classifiers memorise?

2023-10-09 · Michal Lukasik, Vaishnavh Nagarajan, Ankit Singh Rawat, Aditya Krishna Menon, Sanjiv Kumar

The success of modern neural networks has prompted study of the connection between memorisation and generalisation: overparameterised models generalise well, despite being able to perfectly fit (memorise) completely random labels. To carefully study this issue, Feldman proposed a metric to quantify the degree of memorisation of individual training examples, and empirically computed the corresponding memorisation profile of a ResNet on image classification bench-marks. While an exciting first glimpse into what real-world models memorise, this leaves open a fundamental question: do larger neural models memorise more? We present a comprehensive empirical analysis of this question on image classification benchmarks. We find that training examples exhibit an unexpectedly diverse set of memorisation trajectories across model sizes: most samples experience decreased memorisation under larger models, while the rest exhibit cap-shaped or increasing memorisation. We show that various proxies for the Feldman memorization score fail to capture these fundamental trends. Lastly, we find that knowledge distillation, an effective and popular model compression technique, tends to inhibit memorisation, while also improving generalisation. Specifically, memorisation is mostly inhibited on examples with increasing memorisation trajectories, thus pointing at how distillation improves generalisation.

📄 PDF Abstract BibTeX arXiv:2310.05337

Code (0)

등록된 구현이 없습니다.

Tasks

image-classificationImage ClassificationKnowledge DistillationMemorizationModel Compression

Methods 이 논문이 사용한 방법론

ReLU How Do I Communicate to Expedia? How Do I Communicate to Expedia? – Call ☎️ +1-(888) 829 (0881) or +1-805-330-4056 or +1-805-330-4056 for Live Support & Special Travel…
Batch Normalization 설명 없음
1x1 Convolution A 1 x 1 Convolution is a convolution with some special properties in that it can be used for dimensionality reduction,…
Average Pooling 설명 없음
Bottleneck Residual Block A Bottleneck Residual Block is a variant of the residual block that utilises 1x1 convolutions to create a bottleneck. The…
Convolution A convolution is a type of matrix operation, consisting of a kernel, a small matrix of weights, that slides over input data performing element-wise multiplication with the…
Residual Connection 설명 없음
Max Pooling Max Pooling is a pooling operation that calculates the maximum value for patches of a feature map, and uses it to create a downsampled (pooled) feature map. It is usually…

Similar Papers 제목 키워드 기반

Understanding Memorisation in LLMs: Dynamics, Influencing Factors, and Implications

2024-07-27 · Till Speicher, Mohammad Aflah Khan, Qinyuan Wu, Vedant Nanda 외

Understanding whether and to what extent large language models (LLMs) have memorised training data has important implications for the reliability of their output and the privacy of their training data. In order to cleanl…

In-Context Learning

Measuring Unintended Memorisation of Unique Private Features in Neural Networks

2022-02-16 · John Hartley, Sotirios A. Tsaftaris

Neural networks pose a privacy risk to training data due to their propensity to memorise and leak information. Focusing on image classification, we show that neural networks also unintentionally memorise unique features …

image-classificationImage Classification

Unintended memorisation of unique features in neural networks

2022-05-20 · John Hartley, Sotirios A. Tsaftaris

Neural networks pose a privacy risk due to their propensity to memorise and leak training data. We show that unique features occurring only once in training data are memorised by discriminative multi-layer perceptrons an…

Memorisation, convergence and generalisation in generative models

2026-05-20 · Antoine Maillard, Sebastian Goldt arxiv

Generative neural networks learn how to produce highly realistic images from a large, but finite number of examples - or do they simply memorise their training set? To settle this question, Kadkhodaie, Guth, Simoncelli a…

Model Capacity Determines Grokking through Competing Memorisation and Generalisation Speeds

2026-05-10 · Yiding Song, Hanming Ye arxiv

Existing accounts of grokking explain the phenomena in terms of mechanistic frameworks such as circuit efficiency or lazy-to-rich transitions. However, despite a known dependence between grokking and model size, how mode…