ViLT is a faster vision-and-language model without convolution or region supervision.
problem Efficiency and expressive power limitations in current VLP models.
method A convolution-free Vision-and-Language Transformer (ViLT) that processes visual inputs similarly to textual inputs.
result ViLT is up to tens of times faster with competitive or better performance.
Vision and language tasks often fail to test AI comprehensively.
problem Current vision and language tasks are flawed due to dataset and evaluation issues.
method Review of current state and proposal for improvement.
result State-of-the-art systems perform well due to dataset and evaluation flaws.
Paper proposes multiscale self-attentive convolutions for vision and language.
problem Improving language and vision understanding models using self-attention.
method Developed 1D and 2D Self Attentive Convolutions (SAC), multiscale SAC (MSAC).
result MSAC enhances model performance for vision and language tasks.
A new system combines vision and language for person re-identification.
problem Real-world surveillance lacks visual data for person re-identification.
method Two-stream CNN framework with shared logits, CCA for modalities, multi-modal testing protocol.
result 22% improvement in re-identification performance with multi-modal queries.
VALAN is a framework for navigation agents in photo-realistic environments.
problem Developing agents for indoor navigation tasks.
method Deep reinforcement learning with SEED RL architecture.
result VALAN framework can solve a variety of RL problems.
Plex improves model reliability across vision and language tasks.
problem Improving model reliability in diverse decision-making tasks involving uncertainty and adaptation.
method Developed ViT-Plex and T5-Plex pretrained model extensions to evaluate and improve reliability across 40 datasets.
result Plex greatly improves state-of-the-art across reliability tasks, simplifying evaluation and performance.
System discovers new classes from unlabeled data, improving model performance.
problem Handling datapoints outside initial training distribution.
method Develops new classes through semi-supervised learning, using Dataset Reconstruction Accuracy and class learnability.
result Demonstrates improved model quality through automatic class discovery.
LORL learns object-centric representations from vision and language.
problem Learning disentangled, object-centric scene representations from vision and language.
method LORL integrates unsupervised object discovery and segmentation with language input to learn object-centric concepts.
result LORL improves unsupervised object discovery methods and aids downstream tasks.
Dp-CLIP preserves privacy in multimodal AI training.
problem Privacy concerns in multimodal AI, especially in vision-language tasks.
method Differentially private adaptation of CLIP model.
result Dp-CLIP retains accuracy while ensuring privacy.
FAN improves attention weights for better relation emphasis.
problem Learning attention weights for better relation emphasis.
method Introduced a novel center-mass cross entropy loss and a focused attention backbone.
result Focused supervision leads to improved attention distribution and enhanced representation.
Gold Seeker uses reinforcement learning to select actions that maximize information gain.
problem Active information selection for goal-oriented vision-and-language reasoning.
method Reinforcement learning with policy distributions to represent and reduce uncertainty.
result The method outperforms competitors in visual dialog and visual query generation challenges.
Effective theory for Transformer initialization improves model performance.
problem Improving performance of Transformers at initialization.
method Effective-theory analysis of signal propagation in wide and deep Transformers.
result Particular width scalings of initialization and training hyperparameters.
Paper proposes EEIPU, a memoization-aware BO algorithm to reduce hyperparameter tuning costs.
problem High costs in GPU-days for training and fine-tuning language models.
method Memoization-aware Bayesian Optimization (EEIPU) algorithm in tandem with pipeline caching.
result EEIPU produces 103% more hyperparameter candidates and 108% more validation metric improvement.
Sparse linear models improve neural network debuggability.
problem Improving neural network interpretability and debugging.
method Using sparse linear models over learned deep feature representations.
result The approach leads to more debuggable and accurate neural networks.
We describe a mechanism by which artificial neural networks can learn rapid adaptation - the ability to adapt on the fly, with little data, to new tasks - that we call conditionally shifted neurons. We apply this mechanism in the framework of metalearning, where the aim is to replicate some of the flexibility of human …
Deep learning (DL), despite its enormous success in many computer vision and language processing applications, is exceedingly vulnerable to adversarial attacks. We consider the use of DL for radio signal (modulation) classification tasks, and present practical methods for the crafting of white-box and universal black-b…
D-Adaptation automatically sets optimal learning rates without manual tuning.
problem Optimizing learning rates for efficient convergence in machine learning.
method D-Adaptation, which asymptotically achieves optimal learning rates without back-tracking or additional evaluations.
result D-Adaptation automatically matches hand-tuned learning rates across diverse problems.
Fair Mixup improves fairness in classifiers by interpolating between groups.
problem Ensuring fairness in classifiers during training and evaluation.
method Fair Mixup uses interpolation of samples between groups to enforce fairness constraints.
result Fair Mixup ensures better generalization of fairness in various benchmarks.
RotationOut rotates input vectors to regularize neural networks.
problem Reduction of co-adaptation in neural networks.
method Randomly rotates input vectors of the input layer.
result RotationOut reduces co-adaptation better than Dropout.
Apollo improves nonconvex stochastic optimization efficiency.
problem Nonconvex stochastic optimization challenges.
method Adaptive parameter-wise diagonal quasi-Newton method approximating Hessian.
result Significant improvements in convergence speed and generalization over SGD and Adam.
A new algorithm reduces bias and variance in distributionally robust optimization.
problem Distributionally robust optimization with bias and variance issues.
method Prospect, a stochastic gradient-based algorithm that reduces hyperparameter tuning.
result Prospect achieves linear convergence and 2-3x faster convergence on various benchmarks.
MMVAE learns multi-modal data with shared and private latent spaces.
problem Learning useful representations across multiple data modalities.
method Mixture-of-experts variational autoencoder (MMVAE).
result MMVAE satisfies four criteria for multi-modal learning.
Knowledge Distillation (KD) consists of transferring âknowledgeâ from one machine learning model (the teacher) to another (the student). Commonly, the teacher is a high-capacity model with formidable performance, while the student is more compact. By transferring knowledge, one hopes to benefit from the studentâs…
Generates detailed fashion feedback from outfit images.
problem Creating informative and diverse fashion feedback from outfit images.
method Trained deep generative models with visual attention, then improved with Maximum Mutual Information objective function.
result Generated sentences are more diverse and detailed.
This work optimizes alignment and uniformity of features on a hypersphere for better downstream performance.
problem Improving the performance of contrastive representation learning.
method Identifying and optimizing alignment and uniformity of features on a hypersphere.
result Directly optimizing alignment and uniformity leads to comparable or better performance than contrastive learning.
We predict generalization error across model and dataset sizes.
problem Understanding the dependency of neural network generalization error on model and dataset size.
method Model scaling concept applied to construct a functional form of generalization error.
result The constructed functional form accurately predicts generalization error across scales.
Learning from a few examples remains a key challenge in machine learning. Despite recent advances in important domains such as vision and language, the standard supervised deep learning paradigm does not offer a satisfactory solution for learning new concepts rapidly from little data. In this work, we employ ideas from…
PropFair algorithm ensures fair performance in federated learning.
problem Ensuring fair performance in federated learning for diverse clients.
method PropFair, a novel algorithm based on bargaining games, finds proportionally fair solutions.
result PropFair approximately finds proportional fairness solutions and balances average and worst 10% client performances.
DM improves deep model robustness for noisy, imbalanced datasets.
problem Noisy labels and imbalanced datasets in real-world large-scale datasets.
method Derivative Manipulation (DM) approach to example weighting.
result DM enhances robustness of deep models under adverse conditions.
Improved NiNo networks accelerate Adam training by up to 50%.
problem Accelerating neural network training with stable and efficient methods.
method Proposed NiNo networks that leverage neuron connectivity and graph neural networks to nowcast parameters periodically during Adam training.
result Accelerates Adam training by up to 50% in vision and language tasks.
Develops a contrastive framework for data-efficient multimodal learning.
problem Expensive training of multimodal generative models requiring related multimodal data.
method Contrastive framework for multimodal learning, distinguishing related from unrelated data.
result Data-efficient multimodal learning on challenging datasets for various VAE models.
Risk control improves EENNs to make faster predictions without sacrificing accuracy.
problem Determining safe times for EENNs to exit early without degrading performance.
method Adapting risk control frameworks to EENNs to tune their exiting mechanism.
result Risk control enables EENNs to make faster predictions while maintaining user-specified performance goals.
Graph data augmentation improves GNN performance in node classification.
problem Improving generalizability of graph neural networks (GNNs) in semi-supervised node classification.
method Introduces GAug framework for graph data augmentation using neural edge predictors.
result GAug framework improves GNN-based node classification performance across various architectures and datasets.
This work interprets GELU and related activations via a first-order loss function.
problem Understanding and optimizing activation functions in neural networks.
method Complementary interpretation using the Gaussian first-order loss function.
result Calibrated or learned uniform-threshold gates are competitive and often outperform GELU, ReLU, and SiLU/Swish.
SAMPa speeds up SAM by parallelizing its computations.
problem Improving neural network generalization through SAM.
method Parallelizing the two gradient computations in SAM.
result Achieves a twofold speedup of SAM.
LassoFlexNet improves deep learning performance on tabular data.
problem Deep learning underperforms tree-based models on tabular data.
method Incorporates five inductive biases and uses Tied Group Lasso for variable selection.
result LassoFlexNet matches or outperforms leading tree-based models on 52 datasets.
Distributed Lion optimizes large model training by reducing communication costs.
problem Training large AI models efficiently with reduced communication costs.
method Adapted Lion optimizer for distributed training, using binary or lower-precision vectors for communication.
result Distributed Lion achieves comparable performance to standard optimizers but with significantly reduced communication bandwidth.
Bayesian neural networks are vulnerable to adversarial attacks.
problem Adversarial robustness of Bayesian neural networks.
method Examination of adversarial robustness through three tasks: label prediction, adversarial example detection, and semantic shift detection.
result Bayesian neural networks are highly susceptible to adversarial attacks.
Many structured prediction problems (particularly in vision and language domains) are ambiguous, with multiple outputs being correct for an input - e.g. there are many ways of describing an image, multiple ways of translating a sentence; however, exhaustively annotating the applicability of all possible outputs is intr…
Large language models predict human sensory judgments across multiple modalities.
problem Determining the extent of perceptual information in language.
method State-of-the-art large language models were used to predict sensory judgments across six psychophysical datasets.
result Large language models can predict human sensory judgments across multiple modalities with significant correlation to human data.
Enhances graph neural networks by creating virtual data examples.
problem Lack of examples to identify optimal graph rationales in graph applications.
method Introduces environment replacement to create virtual data examples and proposes a framework for rationale-environment separation and representation learning.
result Demonstrates the effectiveness and efficiency of the augmentation-based graph rationalization framework on molecular and polymer datasets.
New method tests black box models for important features.
problem Control false discoveries in high-stakes model interpretations.
method Reframe interpretability as hypothesis testing, propose two testing methods.
result Tests control false discovery rate and select intuitive features.
VNLA uses vision and language to guide agents in finding objects in indoor environments.
problem Guiding agents in finding objects in indoor environments via language.
method Developed I3L framework for imitation learning with indirect intervention.
result Significantly improved success rate of learning agents over baselines.
Our research proves neural collapse in deep ResNets and transformers is globally optimal.
problem Understanding neural collapse in deep learning models.
method Analysis of deep regularized transformers and ResNets trained with cross entropy or mean squared error loss.
result Global optima of deep regularized transformers and ResNets are approximately collapsed, becoming more prominent as depth increases.
Transformers are less sensitive to input perturbations compared to other models.
problem Understanding the inductive biases of transformers and distinguishing them from other architectures.
method Identified token-wise sensitivity as a metric to explain transformers' inductive biases across different data modalities.
result Transformers have lower sensitivity than MLPs, CNNs, ConvMixers, and LSTMs, across vision and language tasks.
Optimal selective classification using likelihood ratios improves model reliability.
problem Enhancing predictive model reliability by allowing uncertain predictions.
method Neyman--Pearson lemma applied to likelihood ratios for optimal selection.
result Neyman--Pearson-informed methods outperform existing baselines under covariate shifts.
Paper develops a theory explaining contrastive pre-training for multimodal AI.
problem Limited theoretical understanding of contrastive pre-training for multi-modal AI.
method Introduces approximate sufficient statistics and Joint Generative Hierarchical Model.
result Near-minimizers of contrastive loss are approximately sufficient, enabling diverse downstream tasks.
New Bayesian method improves Pareto front estimation in multitask finetuning.
problem Efficiently estimating Pareto fronts for multitask finetuning.
method Variational Model Merging using non-Gaussian posteriors.
result More flexible posteriors lead to better Pareto front estimates.