Erhan Bilal

Bo Wen, Chen Wang, Erhan Bilal

ARC-AGI and ARC-AGI-2 measure generalization-through-composition on small color-quantized grids, and their prize competitions make progress on these harder held-out tasks a meaningful proxy for systematic generalization. Recent instruction-first systems translate grids into concise natural-language or DSL rules executed in generate-execute-select loops, yet we lack a principled account of how encodings shape model perception and how to separate instruction errors from execution errors. We hypothesize that modality imposes perceptual bottlenecks -- text flattens 2D structure into 1D tokens while images preserve layout but can introduce patch-size aliasing -- thereby shaping which grid features are reliably perceived. To test this, we isolate perception from reasoning across nine text and image modalities using a weighted set-disagreement metric and a two-stage reasoning pipeline, finding that structured text yields precise coordinates on sparse features, images capture 2D shapes yet are resolution-sensitive, and combining them improves execution (about 8 perception points; about 0.20 median similarity). Overall, aligning representations with transformer inductive biases and enabling cross-validation between text and image yields more accurate instructions and more reliable execution without changing the underlying model.

Deval Mehta, Umar Asif, Tian Hao et al.

This paper presents a novel deep learning enabled, video based analysis framework for assessing the Unified Parkinsons Disease Rating Scale (UPDRS) that can be used in the clinic or at home. We report results from comparing the performance of the framework to that of trained clinicians on a population of 32 Parkinsons disease (PD) patients. In-person clinical assessments by trained neurologists are used as the ground truth for training our framework and for comparing the performance. We find that the standard sit-to-stand activity can be used to evaluate the UPDRS sub-scores of bradykinesia (BRADY) and posture instability and gait disorders (PIGD). For BRADY we find F1-scores of 0.75 using our framework compared to 0.50 for the video based rater clinicians, while for PIGD we find 0.78 for the framework and 0.45 for the video based rater clinicians. We believe our proposed framework has potential to provide clinically acceptable end points of PD in greater granularity without imposing burdens on patients and clinicians, which empowers a variety of use cases such as passive tracking of PD progression in spaces such as nursing homes, in-home self-assessment, and enhanced tele-medicine.

Erhan Bilal

Stochastic gradient descent (SGD) has been the dominant optimization method for training deep neural networks due to its many desirable properties. One of the more remarkable and least understood quality of SGD is that it generalizes relatively well on unseen data even when the neural network has millions of parameters. We hypothesize that in certain cases it is desirable to relax its intrinsic generalization properties and introduce an extension of SGD called deep gradient boosting (DGB). The key idea of DGB is that back-propagated gradients inferred using the chain rule can be viewed as pseudo-residual targets of a gradient boosting problem. Thus at each layer of a neural network the weight update is calculated by solving the corresponding boosting problem using a linear base learner. The resulting weight update formula can also be viewed as a normalization procedure of the data that arrives at each layer during the forward pass. When implemented as a separate input normalization layer (INN) the new architecture shows improved performance on image recognition tasks when compared to the same architecture without normalization layers. As opposed to batch normalization (BN), INN has no learnable parameters however it matches its performance on CIFAR10 and ImageNet classification tasks.

Omer Dror, Boaz Nadler, Erhan Bilal et al.

Consider a regression problem where there is no labeled data and the only observations are the predictions $f_i(x_j)$ of $m$ experts $f_{i}$ over many samples $x_j$. With no knowledge on the accuracy of the experts, is it still possible to accurately estimate the unknown responses $y_{j}$? Can one still detect the least or most accurate experts? In this work we propose a framework to study these questions, based on the assumption that the $m$ experts have uncorrelated deviations from the optimal predictor. Assuming the first two moments of the response are known, we develop methods to detect the best and worst regressors, and derive U-PCR, a novel principal components approach for unsupervised ensemble regression. We provide theoretical support for U-PCR and illustrate its improved accuracy over the ensemble mean and median on a variety of regression problems.