Sanjiv Kumar

CONDITIONAL DATA GENERATION USING LATENT CONTINUOUS CONDITIONAL RANDOM FIELD MODELS

ESTIMATION TECHNIQUES FOR EFFICIENT NEURAL NETWORK INFERENCE PROCESSING

INCREASING SPARSITY TO IMPROVE NEURAL NETWORK EFFICIENCY

Machine Learning Using Four-Bit Binary Data Formats

NEURAL NETWORKS WITH LEARNED AUGMENTED RESIDUAL LAYERS

Controlled Adaptive Optimization

Systems and Methods for Weighted Quantization

Federated Learning with Adaptive Optimization

HARDWARE-FRIENDLY AND PARAMETER-EFFICIENT TUNING OF NEURAL NETWORKS

Fast Speculative Decoding Using Multiple Parallel Drafts

Adaptive Optimization with Improved Convergence

RELATIVE POSITION BIASES IN ATTENTION NEURAL NETWORKS USING FUNCTIONAL INTERPOLATION

Artificial Intelligence Generated Badges for Search

DYNAMIC SELECTION FROM AMONG MULTIPLE CANDIDATE GENERATIVE MODELS WITH DIFFERING COMPUTATIONAL EFFICIENCIES

PERFORMING CLASSIFICATION TASKS USING POST-HOC ESTIMATORS FOR EXPERT DEFERRAL

Tuning approximate nearest neighbor search engines for speed-recall tradeoffs via Lagrange multiplier methods

Systems and methods for communication efficient distributed mean estimation

Systems and Methods for Weighted Quantization

APPROXIMATE K NEAREST NEIGHBORS ON HARDWARE ACCELERATORS

Controlled Adaptive Optimization

Training machine learning models to be robust against label noise

Efficient Training of Embedding Models Using Negative Cache

Spherical random features for polynomial kernels

Multiscale quantization for fast similarity search

Adaptive optimization with improved convergence

Leveraging Redundancy in Attention with Reuse Transformers

ACCOUNTING FOR LONG-TAIL TRAINING DATA THROUGH LOGIT ADJUSTMENT

Kernelized Classifiers in Neural Networks

MULTI-STAGE COMPUTATIONALLY EFFICIENT NEURAL NETWORK INFERENCE

Systems and methods for communication efficient distributed mean estimation

Federated Learning with Only Positive Labels

Systems and Methods for Machine Learning in Hyperbolic Space

Fast adaptive optimization

Federated learning with adaptive optimization

Systems and methods for weighted quantization

PRIVACY PRESERVING MACHINE LEARNING MODEL TRAINING

Sampled softmax with Random Fourier features

Controlled adaptive optimization

Local orthogonal decomposition for maximum inner product search

Multiscale quantization for fast similarity search

Controlled adaptive optimization

Decreasing neural network inference times using softmax approximation

Adaptive optimization with improved convergence

Sparse recovery autoencoder

Systems and methods for evaluating a loss function or a gradient of a loss function via dual decomposition

Efficient inner product operations

Spherical random features for polynomial kernels

Systems and methods for stochastic generative hashing

Quantization-based fast inner product search

Structured orthogonal random features for kernel-based machine learning

Hierarchical quantization for fast inner product search

Systems and methods for communication efficient distributed mean estimation

Systems and Methods for Communication Efficient Distributed Mean Estimation