Optimal transport metric transfers deep network representations efficiently.
problem Efficiently transfer deep network representations for new tasks.
method Use optimal transport to quantify representation similarity and regularize student network.
result Optimal transport distance promotes similarity between teacher and student representations.
Transfer knowledge from multiple sources to improve matrix completion.
problem Matrix completion with noisy data.
method Aggregating singular subspaces information from multiple sources to solve a two-way PCA problem and transform into a low-dimensional linear regression.
result Guaranteed statistical efficiency in transforming the high-dimensional target matrix completion problem.
URL benchmark evaluates uncertainty quantification in pretrained models.
problem Need for reliable uncertainty estimates in transferable pretrained models.
method Proposes URL benchmark to measure transferability of representations and uncertainty estimates.
result Transferable uncertainty quantification remains challenging but not contradictory to traditional goals.
LEEP measures transferability of learned representations efficiently.
problem Evaluating the transferability of learned representations in machine learning.
method LEEP: Log Expected Empirical Prediction, a simple measure requiring one pass through the target data set.
result LEEP predicts transfer and meta-transfer learning performance and convergence speed, outperforming existing measures.
New findings on representation changes in transfer learning.
problem Understanding representation changes in transfer learning.
method Analysis of pre-trained and fine-tuned representations in transfer learning.
result Representation changes mainly occur in the head, but early layers are also important for cross-domain adaptation.
Transfer learning benefits vary in DRL, affecting representation and policy learning.
problem Understanding how much of DRL's sample complexity is due to representation learning.
method Transfer learning experiments comparing performance on related games.
result Benefits of transfer learning are highly variable and non-symmetric across tasks.
The paper tackles transfer learning for growing matrix representations, improving estimation accuracy.
problem Structured matrix estimation under growing ambient dimensions and latent representations.
method Proposes a general transfer framework decomposing target parameters into embedded source components, low-rank innovations, and sparse edits. Develops an anchored alternating projection estimator.
result Establishes deterministic error bounds that separate target noise, representation growth, and source estimation error, yielding improved rates.
Enhances source domain knowledge with target data for transfer learning.
problem Limited data in target domains and rigid model assumptions in transfer learning.
method Transfer learning through Enhanced Sufficient Representation (TESR).
result TESR enhances source domain knowledge with target data, improving transfer learning performance.
Optimal feature transfer identified through bias-variance analysis.
problem Optimizing feature transfer in transfer learning.
method Simple linear model with fine-grained bias-variance decomposition.
result Optimal pretrained feature transform is naturally sparse.
Paper improves sample efficiency of transfer learning in diffusion models.
problem Diffusion models need too much data to train from scratch.
method Assumes shared low-dimensional representation across tasks for improved sample efficiency.
result Sample complexity of target tasks can be reduced with a well-learned representation.
Deep networks, composed of multiple layers of hierarchical distributed representations, tend to learn low-level features in initial layers and transition to high-level features towards final layers. Paradigms such as transfer learning, multi-task learning, and continual learning leverage this notion of generic hierarch…
Knowledge Transfer (KT) techniques tackle the problem of transferring the knowledge from a large and complex neural network into a smaller and faster one. However, existing KT methods are tailored towards classification tasks and they cannot be used efficiently for other representation learning tasks. In this paper a n…
Method transfers feature representation from large to small models using perception coherence.
problem Transfer feature representation from large to small models.
method Defines perception coherence, proposes loss function to minimize.
result Method outperforms or achieves on-par performance compared to strong baseline methods.
In this work, we address the problem of musical timbre transfer, where the goal is to manipulate the timbre of a sound sample from one instrument to match another instrument while preserving other musical content, such as pitch, rhythm, and loudness. In principle, one could apply image-based style transfer techniques t…
Anti-transfer learning prevents misleading representations for speech tasks.
problem Misleading representations learned from orthogonal tasks in speech processing.
method Penalizes similarity between activations of a network and another trained on an orthogonal task.
result Improves classification accuracy and invariance to the orthogonal task.
One question central to Reinforcement Learning is how to learn a feature representation that supports algorithm scaling and re-use of learned information from different tasks. Successor Features approach this problem by learning a feature representation that satisfies a temporal constraint. We present an implementation…
Improved transfer learning with expert models, reducing compute and speeding up performance.
problem Improving sample efficiency and reducing computational requirements for new tasks.
method Training a diverse set of experts using existing label structures and performance proxies to select the relevant expert for each target task.
result Significant speed-up of 2-3 orders of magnitude compared to competing approaches.
Studies show that the representations learned by deep neural networks can be transferred to similar prediction tasks in other domains for which we do not have enough labeled data. However, as we transition to higher layers in the model, the representations become more task-specific and less generalizable. Recent resear…
Many machine intelligence techniques are developed in E-commerce and one of the most essential components is the representation of IDs, including user ID, item ID, product ID, store ID, brand ID, category ID etc. The classical encoding based methods (like one-hot encoding) are inefficient in that it suffers sparsity pr…
Adversarial transfer learning improves stress assessment across users.
problem Transfer learning challenges in physiological biosignals.
method Disentangled nuisance-robust representations using adversarial networks.
result Adversarial framework enhances cross-subjects stress assessment.
Deep neural networks trained on a wide range of datasets demonstrate impressive transferability. Deep features appear general in that they are applicable to many datasets and tasks. Such property is in prevalent use in real-world applications. A neural network pretrained on large datasets, such as ImageNet, can signifi…
CLIP learns joint image-text representations for zero-shot learning.
problem Understanding and improving zero-shot transfer performance in CLIP.
method Formal study of transferrable representation learning and analysis of zero-shot transfer performance.
result Proposes a new CLIP-type approach that outperforms existing methods.
New statistical guarantees for transfer learning with diverse tasks.
problem Statistical guarantees for transfer learning across different tasks.
method Formal analysis of t+1 tasks with shared representation. result Sample complexity reduction for new tasks using shared representation.
To reduce the large computation and storage cost of a deep convolutional neural network, the knowledge distillation based methods have pioneered to transfer the generalization ability of a large (teacher) deep network to a light-weight (student) network. However, these methods mostly focus on transferring the probabili…
New research shows deep models learn sparse features, limiting transfer learning; ensembling improves performance.
problem Sparse feature learning in deep models limits transfer learning performance.
method Developed a theoretical framework and proposed an ensembling strategy to aggregate multiple models.
result Ensembling yields a 9% improvement in transfer accuracy without extra pretraining cost.
Modern deep transfer learning approaches have mainly focused on learning generic feature vectors from one task that are transferable to other tasks, such as word embeddings in language and pretrained convolutional features in vision. However, these approaches usually transfer unary features and largely ignore more stru…
The paper uses thermodynamics to improve machine learning representation quality.
problem Improving the quality of learned representations for transfer learning.
method Formal connection with thermodynamics, iso-classification process, traversing the equilibrium surface.
result Demonstrates how to transfer representations while keeping classification loss constant.
Paper proposes a new speech representation benchmark and model.
problem Lack of benchmarks for comparing speech representations.
method Unsupervised triplet-loss objective for training a universal non-semantic speech representation.
result Proposed representation outperforms other models on benchmark and transfer learning tasks.
Because it is not feasible to collect training data for every language, there is a growing interest in cross-lingual transfer learning. In this paper, we systematically explore zero-shot cross-lingual transfer learning on reading comprehension tasks with a language representation model pre-trained on multi-lingual corp…
Paper proposes a unified time series forecasting model with adaptive transfer.
problem General forecasting models for diverse time series data.
method Unified representations through Decomposed Frequency Learning and adaptive domain-specific features via Time Series Register.
result State-of-the-art forecasting performance on seven real-world benchmarks.
This study investigates how much knowledge from natural images can be transferred to pathology images.
problem Quantifying how much knowledge from natural images can be transferred to pathology images.
method Proposes a framework to quantify knowledge gain by a particular layer, conducts empirical investigation in pathology image centered transfer learning.
result Early layers of deep models can transfer knowledge to pathology image classification tasks.
The objective of transfer reinforcement learning is to generalize from a set of previous tasks to unseen new tasks. In this work, we focus on the transfer scenario where the dynamics among tasks are the same, but their goals differ. Although general value function (Sutton et al., 2011) has been shown to be useful for k…
Often we wish to transfer representational knowledge from one neural network to another. Examples include distilling a large network into a smaller one, transferring knowledge from one sensory modality to a second, or ensembling a collection of models into a single estimator. Knowledge distillation, the standard approa…
Transfer learning aims at building robust prediction models by transferring knowledge gained from one problem to another. In the semantic Web, learning tasks are enhanced with semantic representations. We exploit their semantics to augment transfer learning by dealing with when to transfer with semantic measurements an…
DM improves self-supervised transfer learning by matching target distributions.
problem Improving self-supervised transfer learning performance.
method Distribution Matching (DM) method that drives representation distribution towards a predefined reference distribution.
result DM outperforms existing methods on target classification tasks.
Deconfounds neural network representation similarity metrics to improve consistency and accuracy.
problem Confounding by population structure in similarity metrics like RSA and CKA.
method Covariate adjustment regression to adjust for confounders.
result Improves detection of semantically similar neural networks and consistency in transfer learning.
In a Massive Open Online Course (MOOC), predictive models of student behavior can support multiple aspects of learning, including instructor feedback and timely intervention. Ongoing courses, when the student outcomes are yet unknown, must rely on models trained from the historical data of previously offered courses. I…
A key question in Reinforcement Learning is which representation an agent can learn to efficiently reuse knowledge between different tasks. Recently the Successor Representation was shown to have empirical benefits for transferring knowledge between tasks with shared transition dynamics. This paper presents Model Featu…
New method for one-shot timbre transfer in music.
problem One-shot timbre transfer in music.
method Self-supervised VQ-VAE extension for disentangled representations.
result Outperforms selected baselines on objective metrics.
A method for disentangling text representations without supervision.
problem Challenges in learning disentangled representations of natural language.
method Information-theoretic guidance to induce independent style and content embeddings.
result High quality disentangled representations in terms of content and style preservation.
Paper proposes a method to learn linear regression models using multiple pre-trained models.
problem Learning a linear regression model with limited target data.
method Representation transfer learning method using multiple pre-trained models.
result The method achieves better sample complexity compared to baseline methods.
Transfer learning and data augmentation improve stock classification performance.
problem Challenges in stock classification due to noise and volatility.
method Pre-trained model on S&P500 index features, transfer learning to new models, data augmentation on feature space.
result Augmentation on feature space leads to 20% increase in risk-adjusted returns.
Transfer entropy shows abnormal brain connectivity in depression.
problem Abnormal brain connectivity in depression.
method Transfer entropy analysis on EEG recordings.
result Aberrated dynamics and direction of information between brain centers in depression.
New method improves transfer and robustness of supervised contrastive learning.
problem Class collapse in supervised contrastive learning leads to poor representation quality.
method Adding a weighted class-conditional InfoNCE loss and a class-conditional autoencoder.
result Improves transfer and robustness on 5 standard datasets and 3 worst-group robustness datasets.
Study shows how pretraining robustness transfers to downstream tasks.
problem Understanding how robustness is transferred from pretraining to downstream tasks.
method Theoretical analysis and practical validation of robustness constraints.
result Robustness of a linear predictor on downstream tasks can be constrained by the robustness of its underlying representation.
An important goal of computer vision is to build systems that learn visual representations over time that can be applied to many tasks. In this paper, we investigate a vision-language embedding as a core representation and show that it leads to better cross-task transfer than standard multi-task learning. In particular…
G5 universal GRAPH-BERT learns graph representations across different datasets.
problem Learning graph representations across diverse graph datasets with distinct input and output configurations.
method G5 introduces a pluggable model architecture with input and output components for each graph data source, connected via a unified layer and fusion layer.
result G5 removes obstacles for cross-graph representation learning and transfer, even for sparse data.
Representation learning is a fundamental but challenging problem, especially when the distribution of data is unknown. We propose a new representation learning method, termed Structure Transfer Machine (STM), which enables feature learning process to converge at the representation expectation in a probabilistic way. We…