Related papers: Decomposing the Generalization Gap in PROTAC Activ…
Modern machine learning methods are often overparametrized, allowing adaptation to the data at a fine level. This can seem puzzling; in the worst case, such models do not need to generalize. This puzzle inspired a great amount of work,…
Predictive uncertainty-a model's self awareness regarding its accuracy on an input-is key for both building robust models via training interventions and for test-time applications such as selective classification. We propose a novel…
Inside individual cells, expression of genes is inherently stochastic and manifests as cell-to-cell variability or noise in protein copy numbers. Since proteins half-lives can be comparable to the cell-cycle length, randomness in…
Preference-based alignment methods (e.g., RLHF, DPO) typically optimize a single scalar objective, implicitly averaging over heterogeneous human preferences. In practice, systematic annotator and user-group disagreement makes mean-reward…
Artificial intelligence, trained via machine learning or computational statistics algorithms, holds much promise for the improvement of small molecule drug discovery. However, structure-activity data are high dimensional with low…
PAC-Bayesian is an analysis framework where the training error can be expressed as the weighted average of the hypotheses in the posterior distribution whilst incorporating the prior knowledge. In addition to being a pure generalization…
The purported "black box" nature of neural networks is a barrier to adoption in applications where interpretability is essential. Here we present DeepLIFT (Deep Learning Important FeaTures), a method for decomposing the output prediction of…
Highly capable AI systems could secretly pursue misaligned goals -- what we call "scheming". Because a scheming AI would deliberately try to hide its misaligned goals and actions, measuring and mitigating scheming requires different…
ProBoost, a new boosting algorithm for probabilistic classifiers, is proposed in this work. This algorithm uses the epistemic uncertainty of each training sample to determine the most challenging/uncertain ones; the relevance of these…
Identifying predictive features from high-dimensional datasets is a major task in biomedical research. However, it is difficult to determine the robustness of selected features. Here, we investigate the performance of randomly chosen…
We investigate the generalization and optimization properties of shallow neural-network classifiers trained by gradient descent in the interpolating regime. Specifically, in a realizable scenario where model weights can achieve arbitrarily…
Recent vision language models (VLMs) like CLIP have demonstrated impressive anomaly detection performance under significant distribution shift by utilizing high-level semantic information through text prompts. However, these models often…
Pre-trained vision-language models (VLMs) have shown impressive performance on various downstream tasks by utilizing knowledge learned from large data. In general, the performance of VLMs on target tasks can be further improved by prompt…
Multi-agent LLM committees replicate the same model under different role prompts and aggregate outputs by majority vote, implicitly assuming that agents contribute complementary evidence. We embed each agent's chain-of-thought rationale and…
Out-of-distribution (OOD) detection is the task of identifying inputs that deviate from the training data distribution. This capability is essential for safely deploying deep computer vision models in open-world environments. In this work,…
Recent attacks show that behavioural unlearning of large language models leaves internal traces recoverable by adversarial probes. We characterise where this retention lives and show it can be surgically removed without measurable…
A variety of problems in distributed control involve a networked system of autonomous agents cooperating to carry out some complex task in a decentralized fashion, e.g., orienting a flock of drones, or aggregating data from a network of…
The opacity of massive pretraining corpora in Large Language Models (LLMs) raises significant privacy and copyright concerns, making pretraining data detection a critical challenge. Existing state-of-the-art methods typically rely on token…
Speculative decoding accelerates autoregressive generation by letting a lightweight draft model propose future tokens that a larger target model then verifies in parallel. In practice, however, draft models are usually trained on broad…
We present a novel algorithm to identify potential dispersed signals of new physics in the slew of published LHC results. It employs a random walk algorithm to introduce sets of new particles, dubbed "proto-models", which are tested against…