Ivan Anokhin — accepted AI-conference papers

2026

Learning From the Past with Cascading Eligibility Traces

Tokiniaina Raharison Ralambomihanta, Ivan Anokhin, Roman Pogodin, Samira Ebrahimi Kahou, Jonathan Cornford, Blake Aaron Richards

ICLR 2026poster

Animals often receive information about errors and rewards after significant delays. In some cases these delays are fixed aspects of neural processing or sensory feedback, for example, there is typically a delay of tens to hundreds of milliseconds between motor actions and visual feedback. The stand…

Cited by 0 Source ↗code ↗Scholar ↗

2025

Handling Delay in Real-Time Reinforcement Learning

Ivan Anokhin, Rishav Rishav, Matthew Riemer, Stephen Chung, Irina Rish, Samira Ebrahimi Kahou

ICLR 2025poster

Real-time reinforcement learning (RL) introduces several challenges. First, policies are constrained to a fixed number of actions per second due to hardware limitations. Second, the environment may change while the network is still computing an action, leading to observational delay. The first issue…

Cited by 1 Source ↗PDF ↗code ↗Scholar ↗

2023

Thinker: Learning to Plan and Act

Stephen Chung, Ivan Anokhin, David Krueger

NeurIPS 2023poster

We propose the Thinker algorithm, a novel approach that enables reinforcement learning agents to autonomously interact with and utilize a learned world model. The Thinker algorithm wraps the environment with a world model and introduces new actions designed for interacting with the world model. Thes…

Cited by 8 Source ↗PDF ↗code ↗Scholar ↗

2022

Embedded Ensembles: infinite width limit and operating regimes

Maksim Velikanov, Roman V. Kail, Ivan Anokhin, Roman Vashurin, Maxim Panov, Alexey Zaytsev +1

AISTATS 2022poster

A memory efficient approach to ensembling neural networks is to share most weights among the ensembled models by means of a single reference network. We refer to this strategy as

Cited by 1 Source ↗PDF ↗Scholar ↗

2021

Image Generators With Conditionally-Independent Pixel Synthesis

Ivan Anokhin, Kirill Demochkin, Taras Khakhulin, Gleb Sterkin, Victor Lempitsky, Denis Korzhenkov

CVPR 2021poster

Existing image generator networks rely heavily on spatial convolutions and, optionally, self-attention blocks in order to gradually synthesize images in a coarse-to-fine manner. Here, we present a new architecture for image generators, where the color value at each pixel is computed independently gi…

Cited by 187 PDF ↗code ↗Scholar ↗

2020

High-Resolution Daytime Translation Without Domain Labels

Ivan Anokhin, Pavel Solovev, Denis Korzhenkov, Alexey Kharlamov, Taras Khakhulin, Aleksei Silvestrov +3

CVPR 2020oral

Modeling daytime changes in high resolution photographs, e.g., re-rendering the same scene under different illuminations typical for day, night, or dawn, is a challenging image manipulation task. We present the high-resolution daytime translation (HiDT) model for this task. HiDT combines a generativ…

Cited by 114 PDF ↗code ↗Scholar ↗

2020

Low-loss connection of weight vectors: distribution-based approaches

Ivan Anokhin, Dmitry Yarotsky

ICML 2020poster

Recent research shows that sublevel sets of the loss surfaces of overparameterized networks are connected, exactly or approximately. We describe and compare experimentally a panel of methods used to connect two low-loss points by a low-loss curve on this surface. Our methods vary in accuracy and com…

Cited by 4 Source ↗PDF ↗Scholar ↗