Stanford DAWN project aims to simplify machine learning development.
problem Time-consuming and expensive machine learning development for non-experts.
method Development of systems and tools for end-to-end machine learning applications.
result Simplified machine learning development for non-experts.
Simple model predicts trajectory probabilities.
problem Forecasting trajectories from visual data.
method Spatio-temporal convolutional neural network.
result Achieves results on par with or better than existing methods.
DSA improves sentence embedding by dynamically attending to words.
problem Efficiently capturing the importance of words in sentences for embedding.
method DSA modifies dynamic routing from capsule networks for self-attention in sentences.
result DSA achieves state-of-the-art results in SNLI with fewer parameters.
Paper extends Stanford's equivalence to virtual knots and proves equivalence to Goussarov-Polyak-Viro's n-equivalence.
problem Characterizing finite type invariants in virtual knot theory.
method Using the lower central series of the pure virtual braid group to define an L_n-equivalence.
result L_n-equivalence on virtual string links is equivalent to Goussarov-Polyak-Viro's n-equivalence.
Paper explores BERT's efficiency on SQuAD2.0, freezing layers and using adapters.
problem Improving BERT's efficiency for SQuAD2.0 while maintaining performance.
method Freezing transformer layers, using adapters, and context-aware convolutional filters.
result Context-aware convolutional filters do not improve practical efficiency.
ReF-ER algorithm improved performance in multi-agent reinforcement learning.
problem Improving performance in multi-agent reinforcement learning environments.
method Extended ReF-ER algorithm to include dependencies between agents and modeled environment dynamics.
result ReF-ER MARL outperforms state-of-the-art algorithms in collaborative environments.
This paper improves confidence measurement in deep metric learning models.
problem Measuring confidence in deep metric learning models is challenging.
method Approximates class distributions using Gaussian kernel smoothing and calibrates the confidence metric.
result Improves generalization and robustness of deep metric learning models.
Predicting battery lifespan from early cycles using deep learning.
problem Predicting battery lifespan from the first 100 cycles.
method Used a deep network design including 1-D convolution, LSTMs, and an optional Attention layer.
result Achieved competitive results, with a test MAPE error of 12.5%.
In this article it is proven that if a knot, K, bounds an imbedded grope of class n, then the knot is n/2-trivial in the sense of Gusarov and Stanford. That is, all type n/2 invariants vanish on K. We also give a simple way to construct all knots bounding a grope of a given class. It is further shown that this result i…
We present the multiplicative recurrent neural network as a general model for compositional meaning in language, and evaluate it on the task of fine-grained sentiment analysis. We establish a connection to the previously investigated matrix-space models for compositionality, and show they are special cases of the multi…
Recently Swatee Naik and Theodore Stanford proved that two S-equivalent knots are related by a finite sequence of doubled-delta moves on their knot diagrams. We show that classical S-equivalence is not sufficient to extend their result to ordered links. We define a new algebraic relation on Seifert matrices, called Str…
Study compares machine learning models and BERT on SQuAD dataset.
problem Comparing performance of machine learning models and BERT on SQuAD dataset.
method Comparative analysis of machine learning models and BERT on SQuAD dataset.
result BERT model outperforms other models on SQuAD dataset but requires more time.
Paper proposes scalable multi-label classification for edge devices using CNN.
problem Challenges in deploying multi-label CNN models on edge devices due to high computation and memory requirements.
method Extends existing multi-label classification methods with a single CNN model and multiple loss and accuracy layers.
result Achieves comparable accuracy with 1.8x less MACC operations, 0.97x reduction in latency and 0.5x, 0.84x, 0.97x reduction in size for generated CNN models.
Predicts hospital patient discharge within 24 hours to optimize resource allocation.
problem Improving hospital resource management and patient care by prioritizing discharge.
method Used eight years of EHR data to train models predicting 24-hour discharge.
result Best models achieved AUROC of 0.85 and AUPRC of 0.53, well calibrated.
BiLSTM model improves NER and negation detection in radiological reports.
problem Automating medical information extraction from radiological reports.
method Bi-directional Long Short-Term Memory (BiLSTM) neural network architecture.
result BiLSTM outperforms traditional rule-based systems for NER and negation detection.
The study proposes methods to estimate personalized treatment effects from high-dimensional observational data.
problem Leveraging electronic medical records for personalized treatment recommendations.
method Three methods for estimating heterogeneous treatment effects using observational data.
result The methods perform well in simulations and on real data from a large randomized trial.
Paper connects volumes of moduli spaces of super Riemann surfaces to integrals over stable Riemann surfaces.
problem Relating volumes of moduli spaces of super Riemann surfaces to integrals over stable Riemann surfaces.
method Relates volumes of moduli spaces of super Riemann surfaces to integrals over the moduli space of stable Riemann surfaces Mg,n. result Proves recursion between volumes of moduli spaces of super hyperbolic surfaces using algebraic geometry.
A new method uses optimal transport for semi-supervised classification.
problem Semi-supervised learning with limited labeled data.
method Optimal transport formulation with Sinkhorn iteration for label assignment.
result Improved performance on CIFAR-10, CIFAR-100, and SVHN datasets compared to FixMatch.
CF-VAE models capture multi-modal distributions for better structured sequence prediction.
problem Challenges in capturing multi-modality of future states in latent variable models.
method Conditional Flow Variational Autoencoders (CF-VAE) with conditional normalizing flows.
result CF-VAE achieves state-of-the-art results on multi-modal structured sequence prediction datasets.
Paper improves image retrieval quality using nonlinear rank approximations.
problem Improving image retrieval quality in high-dimensional feature spaces.
method Computes normalized approximated ranks, converts to similarities, and uses them in a new loss function.
result Significant improvement in image retrieval quality on multiple datasets.
Improved text embeddings enhance retrieval from a knowledge base.
problem Efficiently retrieving relevant paragraphs from a large knowledge base.
method Used Stanford Question Answering Dataset (SQuAD) for open-domain question answering. Compared various text-embedding methods and trained deep residual neural models for retrieval.
result Training deep residual neural models for retrieval purposes significantly improves paragraph recall.
Gradient boosting enhances existing Mendelian models for genetic disease risk prediction.
problem Improving existing Mendelian models for genetic disease risk prediction.
method Combining gradient boosting with existing Mendelian models.
result Improved model outperforms both original and gradient boosting-only models.
Study evaluates how well question-answering models generalize to new data types.
problem Generalization of question-answering models to new data types.
method Constructed new test sets from different domains and evaluated models' performance.
result Models show significant performance drops when tested on new data types.
Scene parsing is an important and challenging prob- lem in computer vision. It requires labeling each pixel in an image with the category it belongs to. Tradition- ally, it has been approached with hand-engineered features from color information in images. Recently convolutional neural networks (CNNs), which automatica…
Hierarchical interpretations explain neural network predictions.
problem Inability to visualize complex, non-linear relationships learned by neural networks.
method Agglomerative contextual decomposition (ACD) for feature clustering and prediction explanation.
result ACD identifies clusters of features predictive to DNN predictions and diagnoses incorrect predictions.
Efficiently poisons offline RLHF models by flipping preference labels.
problem Vulnerability of offline RLHF models to preference label flipping attacks.
method Developed two attack methods: BAL-A and BMP-A, solving a structured binary sparse approximation problem.
result Demonstrated that flipping one preference label induces a parameter-independent shift in the DPO gradient, enabling structured binary sparse approximation.
OGB provides diverse graph datasets for robust ML research.
problem Challenges in scalable and robust graph machine learning.
method Unified evaluation protocol, diverse datasets, and automated pipeline.
result Significant scalability and generalization challenges identified.
New Convolutional Unit improves Batch Whitening performance.
problem Improving the efficiency and effectiveness of Batch Whitening.
method Proposes a new Convolutional Unit that aligns with Batch Whitening theory and empirically analyzes the original Convolutional Unit.
result Significantly improved performance on multiple image classification datasets.
Hierarchical Foresight improves robot vision tasks by planning long-term goals.
problem Compounding uncertainty and scalability issues in long horizon video prediction.
method Subgoal generation and planning using hierarchical visual foresight (HVF).
result Achieves nearly 200% performance improvement in vision-based manipulation tasks.
PFE embeds images into sparse regions for better segmentation.
problem Image segmentation challenges with slowly varying signals and sparse region boundaries.
method Piecewise Flat Embedding (PFE) using sparse signal recovery theory, L1,p regularization, and Bregman iterations.
result PFE enhances image segmentation performance on multiple datasets.
The paper explores how information theory aids in statistical learning models.
problem Characterizing fundamental performance limits in statistical learning models.
method Introduces divergence measures and evidence lower bound (ELBO) in model training.
result Provides a systematic derivation for generative diffusion models.
GCN and GPCA are mathematically connected, leading to improved node classification performance.
problem Improving node classification performance in semi-supervised settings.
method Established a mathematical connection between GCN and GPCA, demonstrating their equivalence and using this to design an effective initialization strategy.
result GPCA paired with a simple MLP achieves similar or better performance than GCN on semi-supervised node classification tasks.
The study shows removing fully connected output layers improves efficiency without sacrificing performance.
problem Large number of parameters in fully connected layers for high-category datasets.
method Examined architectures replacing fully connected output layers with fixed layers and compared performance.
result Fixed classifiers offer no additional benefit over removing the output layer and its parameters.
ProtoPNet uses deep learning to classify images by identifying prototypical parts.
problem Challenging image classification tasks where understanding reasoning is important.
method ProtoPNet architecture that reasons by finding prototypical parts and combining evidence.
result ProtoPNet achieves comparable accuracy to non-interpretable models and provides interpretability.
Paper compares DML methods fairly, finds older ones perform better.
problem Fair comparison of deep metric learning methods.
method Unified implementation of 12 DML algorithms under same conditions.
result Several older DML methods perform significantly better than previously believed.
End-to-end multi-object tracking learns object interactions.
problem Object tracking ignores interactions between objects.
method End-to-end relational reasoning model MOHART.
result Relational reasoning improves tracking and prediction.
Paper uses JIVE to decompose word embeddings, improving sentiment analysis performance.
problem Improving sentiment analysis performance on word embeddings.
method Joint and individual variance explained (JIVE) method for decomposition.
result Mapping word embeddings into joint components improves sentiment analysis performance.
Super-convergence allows neural nets to train faster with large learning rates.
problem Training neural networks too slowly.
method Training with large learning rates and one learning rate cycle.
result Neural networks can be trained an order of magnitude faster.
A new framework solves the causal frame problem using potential levels.
problem How to make decisions based on relevant information without considering irrelevant details.
method Introducing Potential Level (PL) and proposing a PL-based Inference Framework (PLIF).
result PLIF is consistent with causal judgment findings and makes testable predictions.
Neural networks improve clinical note representations for predictive modeling.
problem Challenges in using clinical notes for machine learning due to high dimensionality, sparsity, and scarcity of labeled data.
method Used neural networks and transfer learning to learn representations of clinical notes.
result Neural network representations significantly outperformed baseline representations in predictive models.
AutoAugment learns optimal data augmentation policies automatically.
problem Improving image classifier accuracy through better data augmentation.
method AutoAugment uses a search algorithm to find the best augmentation policies in a defined search space.
result AutoAugment achieves state-of-the-art accuracy on multiple datasets.
This work reveals how label noise can cause a final ascent in neural network performance curves.
problem The impact of label noise on the performance of neural networks.
method Theoretical analysis and extensive experiments on various neural network architectures.
result Label noise can lead to a final ascent in the test loss curve, improving generalization at intermediate model widths.
Deep learning predicts RNA degradation from crowdsourced data.
problem Predicting RNA degradation to improve thermostability.
method Crowdsourced machine learning competition on Kaggle.
result 41% of predictions matched experimental data, and models generalized to longer RNA molecules.
Selective joint fine-tuning improves deep learning with limited data.
problem Insufficient labeled training data for deep learning tasks.
method Joint fine-tuning of shared convolutional layers between source and target tasks using selected training images.
result Improves classification accuracy by 2% - 10% on multiple visual classification tasks.
WILDS 2.0 expands benchmark datasets for unsupervised adaptation.
problem Leveraging unlabeled data for distribution shifts in real-world applications.
method Curated unlabeled data across various applications, tasks, and modalities.
result State-of-the-art methods perform poorly on WILDS datasets.
OLÉ simplifies deep learning by enforcing class orthogonality.
problem Training deep networks for image classification without enforcing intra-class similarity and inter-class margin.
method OLÉ collapses class features into a learned subspace and pushes subspaces to be orthogonal.
result OLÉ improves classification performance and robustness.
Bayesian segmentation and uncertainty estimation improve 3D model accuracy for factory planning.
problem Generating accurate 3D models from outdated and incomplete 2D data.
method Bayesian neural network for point cloud segmentation and entropy-based uncertainty estimation.
result Bayesian segmentation network significantly improves model accuracy and object identification.
Deep learning improves PS pixel selection in SAR interferometry.
problem Selecting persistent scatterer pixels for geophysical parameter estimation in multi-temporal SAR interferometry.
method Proposed two deep learning architectures: CNN-ISS and CLSTM-ISS trained on phase history to classify PS and non-PS pixels.
result CLSTM-ISS outperforms conventional methods in PS pixel selection and classification accuracy.