Vinu Joseph

Charles Gouert, Vinu Joseph, Steven Dalton et al.

Fully Homomorphic Encryption (FHE) is a cryptographic method that guarantees the privacy and security of user data during computation. FHE algorithms can perform unlimited arithmetic computations directly on encrypted data without decrypting it. Thus, even when processed by untrusted systems, confidential data is never exposed. In this work, we develop new techniques for accelerated encrypted execution and demonstrate the significant performance advantages of our approach. Our current focus is the Fully Homomorphic Encryption over the Torus (CGGI) scheme, which is a current state-of-the-art method for evaluating arbitrary functions in the encrypted domain. CGGI represents a computation as a graph of homomorphic logic gates and each individual bit of the plaintext is transformed into a polynomial in the encrypted domain. Arithmetic on such data becomes very expensive: operations on bits become operations on entire polynomials. Therefore, evaluating even relatively simple nonlinear functions, such as a sigmoid, can take thousands of seconds on a single CPU thread. Using our novel framework for end-to-end accelerated encrypted execution called ArctyrEX, developers with no knowledge of complex FHE libraries can simply describe their computation as a C program that is evaluated over $40\times$ faster on an NVIDIA DGX A100 and $6\times$ faster with a single A100 relative to a 256-threaded CPU baseline.

Harvey Dam, Vinu Joseph, Aditya Bhaskara et al.

Network compression is now a mature sub-field of neural network research: over the last decade, significant progress has been made towards reducing the size of models and speeding up inference, while maintaining the classification accuracy. However, many works have observed that focusing on just the overall accuracy can be misguided. E.g., it has been shown that mismatches between the full and compressed models can be biased towards under-represented classes. This raises the important research question, can we achieve network compression while maintaining "semantic equivalence" with the original network? In this work, we study this question in the context of the "long tail" phenomenon in computer vision datasets observed by Feldman, et al. They argue that memorization of certain inputs (appropriately defined) is essential to achieving good generalization. As compression limits the capacity of a network (and hence also its ability to memorize), we study the question: are mismatches between the full and compressed models correlated with the memorized training data? We present positive evidence in this direction for image classification tasks, by considering different base architectures and compression schemes.

Vinu Joseph, Saurav Muralidharan, Animesh Garg et al.

Deep neural networks (DNNs) frequently contain far more weights, represented at a higher precision, than are required for the specific task which they are trained to perform. Consequently, they can often be compressed using techniques such as weight pruning and quantization that reduce both the model size and inference time without appreciable loss in accuracy. However, finding the best compression strategy and corresponding target sparsity for a given DNN, hardware platform, and optimization objective currently requires expensive, frequently manual, trial-and-error experimentation. In this paper, we introduce a programmable system for model compression called Condensa. Users programmatically compose simple operators, in Python, to build more complex and practically interesting compression strategies. Given a strategy and user-provided objective (such as minimization of running time), Condensa uses a novel Bayesian optimization-based algorithm to automatically infer desirable sparsities. Our experiments on four real-world DNNs demonstrate memory footprint and hardware runtime throughput improvements of 188x and 2.59x, respectively, using at most ten samples per search. We have released a reference implementation of Condensa at https://github.com/NVlabs/condensa.

Harvey Dam, Jonas Knochelmann, Vinu Joseph et al.

We introduce a method to reduce refusal rates of large language models (LLMs) on sensitive content without modifying model weights or prompts. Motivated by the observation that refusals in certain models were often preceded by the specific token sequence of a token marking the beginning of the chain-of-thought (CoT) block (<think>) followed by a double newline token (\n\n), we investigate the impact of two simple formatting adjustments during generation: suppressing \n\n after <think> and suppressing the end-of-sequence token after the end of the CoT block (</think>). Our method requires no datasets, parameter changes, or training, relying solely on modifying token probabilities during generation. In our experiments with official DeepSeek-R1 distillations, these interventions increased the proportion of substantive answers to sensitive prompts without affecting performance on standard benchmarks. Our findings suggest that refusal behaviors can be circumvented by blocking refusal subspaces at specific points in the generation process.

Vinu Joseph, Shoaib Ahmed Siddiqui, Aditya Bhaskara et al.

With the rise in edge-computing devices, there has been an increasing demand to deploy energy and resource-efficient models. A large body of research has been devoted to developing methods that can reduce the size of the model considerably without affecting the standard metrics such as top-1 accuracy. However, these pruning approaches tend to result in a significant mismatch in other metrics such as fairness across classes and explainability. To combat such misalignment, we propose a novel multi-part loss function inspired by the knowledge-distillation literature. Through extensive experiments, we demonstrate the effectiveness of our approach across different compression algorithms, architectures, tasks as well as datasets. In particular, we obtain up to $4.1\times$ reduction in the number of prediction mismatches between the compressed and reference models, and up to $5.7\times$ in cases where the reference model makes the correct prediction; all while making no changes to the compression algorithm, and minor modifications to the loss function. Furthermore, we demonstrate how inducing simple alignment between the predictions of the models naturally improves the alignment on other metrics including fairness and attributions. Our framework can thus serve as a simple plug-and-play component for compression algorithms in the future.

Mark Van der Merwe, Vinu Joseph, Ganesh Gopalakrishnan

Belief Propagation (BP) is a message-passing algorithm for approximate inference over Probabilistic Graphical Models (PGMs), finding many applications such as computer vision, error-correcting codes, and protein-folding. While general, the convergence and speed of the algorithm has limited its practical use on difficult inference problems. As an algorithm that is highly amenable to parallelization, many-core Graphical Processing Units (GPUs) could significantly improve BP performance. Improving BP through many-core systems is non-trivial: the scheduling of messages in the algorithm strongly affects performance. We present a study of message scheduling for BP on GPUs. We demonstrate that BP exhibits a tradeoff between speed and convergence based on parallelism and show that existing message schedulings are not able to utilize this tradeoff. To this end, we present a novel randomized message scheduling approach, Randomized BP (RnBP), which outperforms existing methods on the GPU.