Shreyank N Gowda

Collateral damage in concept erasure is significantly reduced, allowing for precise edits without compromising related visual information.

Probabilistic forecasting of Alzheimer's progression reveals that uncertainty grows significantly over time, especially for patients with rare progression patterns.

QC-SMOTE outperforms traditional methods by ensuring synthetic samples are generated with high reliability, adapting to the local data structure to avoid noise and overlap pitfalls.

Zero-shot financial NLP fails to beat simple baselines, revealing critical limitations in predicting stock movements from news sentiment.

Explanation quality can be harnessed to filter out noise in ECG classification, leading to significant efficiency gains in model training.

Prompt injection detection performance varies wildly across deployment settings, so relying on leaderboard rankings alone could leave your LLM vulnerable.

Current open-world semi-supervised learning methods fall short in practical applications because they fail to extract latent semantic information, but SECOS overcomes this by directly predicting textual labels from a candidate set, achieving state-of-the-art results.

Training deep nets doesn't need to be a data deluge: dynamically dropping less-useful training examples during learning can maintain accuracy while slashing compute.

Image compression, a seemingly benign process, can drastically amplify the power of adversarial attacks, making your image classifiers far more vulnerable than you thought.