Mehrab Hamidi

Karolis Jucys, George Adamopoulos, Mehrab Hamidi et al.

Understanding the mechanisms behind decisions taken by large foundation models in sequential decision making tasks is critical to ensuring that such systems operate transparently and safely. In this work, we perform exploratory analysis on the Video PreTraining (VPT) Minecraft playing agent, one of the largest open-source vision-based agents. We aim to illuminate its reasoning mechanisms by applying various interpretability techniques. First, we analyze the attention mechanism while the agent solves its training task - crafting a diamond pickaxe. The agent pays attention to the last four frames and several key-frames further back in its six-second memory. This is a possible mechanism for maintaining coherence in a task that takes 3-10 minutes, despite the short memory span. Secondly, we perform various interventions, which help us uncover a worrying case of goal misgeneralization: VPT mistakenly identifies a villager wearing brown clothes as a tree trunk when the villager is positioned stationary under green tree leaves, and punches it to death.

Simon Dufort-Labbé, Mehrab Hamidi, Razvan Pascanu et al.

Sharpness-aware minimization (SAM) encourages flat minima by perturbing parameters along directions of high loss curvature, but treats all parameter directions uniformly, ignoring the underlying loss geometry. We introduce LLQR+SAM, which combines SAM with a learned preconditioner obtained from the recently proposed LLQR framework, a second-order method that recasts steepest descent as a layerwise linear-quadratic regulator problem. The preconditioner is updated sparsely and maintained as a slow exponential moving average, so it captures a smoothed, low-resolution picture of the loss landscape geometry. The SAM perturbation then operates on top of this learned geometry, probing curvature at a faster timescale. We show that this two-timescale structure is not merely a computational convenience: theoretically, the preconditioner amplifies the SAM escape signal in directions that are flat under the average geometry but locally sharp (potholes). Wide, flat basins, by contrast, remain stable. Empirically, LLQR+SAM gives consistent gains over both SAM and LLQR alone across standard vision and sequence modeling benchmarks, supporting the view that slow learned geometry and fast sharpness correction are genuinely complementary.

Alireza Dizaji, Benedict Aaron Tjandra, Mehrab Hamidi et al.

Dynamic graph learning methods have recently emerged as powerful tools for modelling relational data evolving through time. However, despite extensive benchmarking efforts, it remains unclear whether current Temporal Graph Neural Networks (TGNNs) effectively capture core temporal patterns such as periodicity, cause-and-effect, and long-range dependencies. In this work, we introduce the Temporal Graph Reasoning Benchmark (T-GRAB), a comprehensive set of synthetic tasks designed to systematically probe the capabilities of TGNNs to reason across time. T-GRAB provides controlled, interpretable tasks that isolate key temporal skills: counting/memorizing periodic repetitions, inferring delayed causal effects, and capturing long-range dependencies over both spatial and temporal dimensions. We evaluate 11 temporal graph learning methods on these tasks, revealing fundamental shortcomings in their ability to generalize temporal patterns. Our findings offer actionable insights into the limitations of current models, highlight challenges hidden by traditional real-world benchmarks, and motivate the development of architectures with stronger temporal reasoning abilities. The code for T-GRAB can be found at: https://github.com/alirezadizaji/T-GRAB.

Mehrab Hamidi

Reverse engineering deep ReLU networks is a critical problem in understanding the complex behavior and interpretability of neural networks. In this research, we present a novel method for reconstructing deep ReLU networks by leveraging convex optimization techniques and a sampling-based approach. Our method begins by sampling points in the input space and querying the black box model to obtain the corresponding hyperplanes. We then define a convex optimization problem with carefully chosen constraints and conditions to guarantee its convexity. The objective function is designed to minimize the discrepancy between the reconstructed networks output and the target models output, subject to the constraints. We employ gradient descent to optimize the objective function, incorporating L1 or L2 regularization as needed to encourage sparse or smooth solutions. Our research contributes to the growing body of work on reverse engineering deep ReLU networks and paves the way for new advancements in neural network interpretability and security.

Rassa Ghavami Modegh, Mehrab Hamidi, Saeed Masoudian et al.

COVID-19 is a virus with high transmission rate that demands rapid identification of the infected patients to reduce the spread of the disease. The current gold-standard test, Reverse-Transcription Polymerase Chain Reaction (RT-PCR), has a high rate of false negatives. Diagnosing from CT-scan images as a more accurate alternative has the challenge of distinguishing COVID-19 from other pneumonia diseases. Artificial intelligence can help radiologists and physicians to accelerate the process of diagnosis, increase its accuracy, and measure the severity of the disease. We designed a new interpretable deep neural network to distinguish healthy people, patients with COVID-19, and patients with other pneumonia diseases from axial lung CT-scan images. Our model also detects the infected areas and calculates the percentage of the infected lung volume. We first preprocessed the images to eliminate the batch effects of different devices, and then adopted a weakly supervised method to train the model without having any tags for the infected parts. We trained and evaluated the model on a large dataset of 3359 samples from 6 different medical centers. The model reached sensitivities of 97.75% and 98.15%, and specificities of 87% and 81.03% in separating healthy people from the diseased and COVID-19 from other diseases, respectively. It also demonstrated similar performance for 1435 samples from 6 different medical centers which proves its generalizability. The performance of the model on a large diverse dataset, its generalizability, and interpretability makes it suitable to be used as a reliable diagnostic system.