No abstract

The 3D Bin Packing Problem (3D-BPP) is one of the most demanded yet challenging problems in industry, where an agent must pack variable size items delivered in sequence into a finite bin with the aim to maximize the space utilization. It represents a strongly NP-Hard optimization problem such that no solution has been offered to date with high performance in space utilization. In this paper, we present a new reinforcement learning (RL) framework for a 3D-BPP solution for improving performance. First, a buffer is introduced to allow multi-item action selection. By increasing the degree of freedom in action selection, a more complex policy that results in better packing performance can be derived. Second, we propose an agnostic data augmentation strategy that exploits both bin item symmetries for improving sample efficiency. Third, we implement a model-based RL method adapted from the popular algorithm AlphaGo, which has shown superhuman performance in zerosum games. Our adaptation is capable of working in singleplayer and score based environments. In spite of the fact that AlphaGo versions are known to be computationally heavy, we manage to train the proposed framework with a single thread and GPU, while obtaining a solution that outperforms the stateof-the-art results in space utilization.

Cell detection is a fundamental task in computational pathology that can be used for extracting high-level medical information from whole-slide images. For accurate cell detection, pathologists often zoom out to understand the tissuelevel structures and zoom in to classify cells based on their morphology and the surrounding context. However, there is a lack of efforts to reflect such behaviors by pathologists in the cell detection models, mainly due to the lack of datasets containing both cell and tissue annotations with overlapping regions. To overcome this limitation, we propose and publicly release OCELOT, a dataset purposely dedicated to the study of cell-tissue relationships for cell detection in histopathology. OCELOT provides overlapping cell and tissue annotations on images acquired from multiple organs. Within this setting, we also propose multi-task learning approaches that benefit from learning both cell and tissue tasks simultaneously. When compared against a model trained only for the cell detection task, our proposed approaches improve cell detection performance on 3 datasets: proposed OCELOT, public TIGER, and internal CARP datasets. On the OCELOT test set in particular, we show up to 6.79 improvement in F1-score. We believe the contributions of this paper, including the release of the OCELOT dataset at https://lunit-io.github.io/research/ publications/ocelot are a crucial starting point toward the important research direction of incorporating celltissue relationships in computation pathology.

e13546 Background: Programmed death-ligand 1 (PD-L1) is a predictive marker for immune checkpoint inhibitors treatment response in urothelial carcinoma (UC). The combined positive score (CPS) is a representative method to evaluate the expression level of PD-L1 in UC. However, inter-observer and inter-institute variations can disrupt accurate CPS evaluation. The purpose of this study is to assess the role of an artificial intelligence (AI)-powered PD-L1 CPS analyzer on UC in reducing inter-observer and inter-institute variability. Methods: Lunit SCOPE PD-L1 CPS was developed with 4.94 x 105 tumor cells and 4.17 x 105 immune cells from 360 PD-L1 stained whole-slide images (WSIs) of UC from multiple institutions. The algorithm consisted of tissue area segmentation and cell detection AI models. The AI models calculated the CPS by detecting tumor cells over the tumor area and immune cells over the tumor and adjacent area. Three uropathologists from different university hospitals evaluated the CPS classification (≥10 or <10) of 543 PD-L1 stained WSIs of UC from each hospital. The result with concordant CPS classification of each slide across ≥ 2 pathologists was considered the consensus. Each pathologist revisited to evaluate WSIs by referencing the AI model inference, after a washout period, if there was a discrepancy between the pathologist and the AI model. Results: Of 543 WSIs, 446 (82.1%) were classified as the same CPS subgroup by all three uropathologists. Also, pathologists had a high degree of concordance with the consensus in WSIs from their own hospitals. They re-evaluated 64, 73, and 75 WSIs with AI assistance, respectively, and changed the CPS classification for 47, 48, and 48 WSIs. After re-evaluation with AI assistance, three uropathologists agreed on the same CPS classification in 510 WSIs (93.9%). The overall percentage agreement (OPA) of each pathologist with the consensus increased from 95.0%, 94.8%, and 92.3% to 98.7%, 98.3%, and 96.9% after AI assistance, and the OPA for WSIs of other institutions increased more compared to the OPA for WSIs of their own hospital. Conclusions: This study shows that an AI-powered PD-L1 CPS analyzer in UC can reduce inter-observer and inter-site variability. This result suggests that the AI model will help evaluate CPS in UC more accurately and reduce variation in situations where pathologists analyze WSIs from unfamiliar institutions.[Table: see text]

In this paper, we present a novel audio synthesizer, CAESynth, based on a conditional autoencoder. CAESynth synthesizes timbre in real-time by interpolating the reference sounds in their shared latent feature space, while controlling a pitch independently. We show that training a conditional autoencoder based on accuracy in timbre classification together with adversarial regularization of pitch content allows timbre distribution in latent space to be more effective and stable for timbre interpolation and pitch conditioning. The proposed method is applicable not only to creation of musical cues but also to exploration of audio affordance in mixed reality based on novel timbre mixtures with environmental sounds. We demonstrate by experiments that CAESynth achieves smooth and high-fidelity audio synthesis in real-time through timbre interpolation and independent yet accurate pitch control for musical cues as well as for audio affordance with environmental sound. A Python implementation along with some generated samples are shared online.