No abstract

No abstract

AI-synthesized face-swapping videos, commonly known as DeepFakes, is an emerging problem threatening the trustworthiness of online information. The need to develop and evaluate DeepFake detection algorithms calls for large-scale datasets. However, current DeepFake datasets suffer from low visual quality and do not resemble Deep-Fake videos circulated on the Internet. We present a new large-scale challenging DeepFake video dataset, Celeb-DF 1 , which contains 5, 639 high-quality DeepFake videos of celebrities generated using improved synthesis process. We conduct a comprehensive evaluation of DeepFake detection methods and datasets to demonstrate the escalated level of challenges posed by Celeb-DF.

Stereoselective total syntheses of the racemic form and the natural enantiomer of the tricyclic marine alkaloid lepadiformine (6) have been accomplished using a novel intramolecular spirocyclization of an N-acyliminium ion with an allylsilane to form the A/C rings as the key step. Introduction of the hydroxymethyl group at C-13 of the racemic spirocycle 11 was achieved using our methodology for oxidative radical-based remote functionalization of o-aminobenzamides, followed by copper-catalyzed addition of Grignard reagent 16 to the N-acyliminium ion intermediate derived from 15. Subsequent Tamao oxidation of silane 17 then afforded the requisite hydroxymethyl compound 19, which was converted to the dimethyl acetal 25 via hydroformylation followed by aldehyde protection. Hydrolysis of the benzamide moiety of 25 and subsequent protection of the primary alcohol gave amino acetal 27. The synthesis was concluded from 27 by a four-step procedure: acid-catalyzed ring closure, amino nitrile formation, introduction of the hexyl chain by a Grignard reaction to an iminium salt, and removal of the O-benzyl protecting group to give (+/-)-lepadiformine (6). The enantioselective total synthesis of 6 started from known optically pure bromide 37, derived from (S)-pyroglutamic acid, and followed a similar sequence involving the key spirocyclization of N-acyliminium ion 42. This synthesis has established the absolute configuration of naturally occurring lepadiformine to be 2(R),5(S),10(S),13(S).

A Design of Experiment (DoE) and kinetic screening study was carried out using an automated reaction screening platform, and as a case study, an early stage in the synthesis of a late phase developmental candidate was investigated. Key impurities were tracked and kinetically modeled, and significant factors impacting impurity formation were identified. In particular, factors that influence the formation of the diastereomer 4, a precursor to an API impurity identified as a Critical Quality Attribute (CQA), were identified and optimized to minimize its formation. Acetic acid, methanesulfonic acid, volumes of solvent, amino alcohol, and reaction B temperature were observed to be the most significant factors along with a factor interaction between methanesulfonic acid and the reaction B temperature. From the experimental data, diastereomer levels of 2.5–5.4 mol % were observed and a kinetic model was developed around the diastereomer formation. Good agreement between the model and experimental data gave confidence in understanding the contributing factors of diastereomer generation, and enabled confirmation of process parameter recommendations to support risk assessments and Quality by Design (QbD) activities. In total, automation provided a 4–5 times savings in FTE hours over a manual process when conducting these experiments and greatly accelerated the generation of supporting information for a drug file.