Explore

Slice the corpus on any axis.

Filter by firm, year, function, metric or slide category — then see the result as a narrative arc, a treemap of slide types, or a catalog of slides.

clear
51,416 slides across 1229 decks match.
Treemap Arc Catalog
Institutions commonly use privates in their allocations
UBS · 2023 · market_landscape
Institutions commonly use privates in their allocations
30% — Asset allocation
 How much should I allocate to private markets?
UBS · 2023 · framework_other
How much should I allocate to private markets?
 Appraisal based pricing can understate risk
UBS · 2023 · diagnosis
Appraisal based pricing can understate risk
23.8% — Volatility
 Adding alternatives can improve a portfolio’s risk / return profile
UBS · 2023 · scatter_plot
Adding alternatives can improve a portfolio’s risk / return profile
 Manager selection is key
UBS · 2023 · market_landscape
Manager selection is key
31.0% — Average annual manager returns
 Committing across vintage years mitigates entry point risks
UBS · 2023 · analyze_data
Committing across vintage years mitigates entry point risks
 J-Curve Effect: Capital Calls early in fund life, followed by distributions near the end
UBS · 2023 · framework_other
J-Curve Effect: Capital Calls early in fund life, followed by distributions near the end
 Achieving target allocation through multi-year commitments
UBS · 2023 · financial_analysis
Achieving target allocation through multi-year commitments
25% — Portfolio NAV
 Managing liquidity and maintenance
UBS · 2023 · financial_analysis
Managing liquidity and maintenance
21% — Portfolio NAV
 Private markets are expected to provide a higher return than public markets
UBS · 2023 · market_landscape
Private markets are expected to provide a higher return than public markets
10.3% — Strategic compound return
 Reported returns for private markets tend to exhibit smoothing
UBS · 2023 · analyze_data
Reported returns for private markets tend to exhibit smoothing
38.6% — Return volatility
 Private markets are gaining importance for UHNW and family office investors
UBS · 2023 · market_landscape
Private markets are gaining importance for UHNW and family office investors
43% — Asset allocation percentage
 Scorecard: Reviewing our predictions from 2018
AirStreetCapital · section_divider
Scorecard: Reviewing our predictions from 2018
 Reinforcement learning (RL) conquers new territory: Montezuma's Revenge
AirStreetCapital · case_study
Reinforcement learning (RL) conquers new territory: Montezuma's Revenge
 RL conquers new territory: StarCraft II
AirStreetCapital · case_study
RL conquers new territory: StarCraft II
$26M — Estimated MMR
 RL conquers new territory: Quake III Arena Capture the Flag
AirStreetCapital · case_study
RL conquers new territory: Quake III Arena Capture the Flag
1500 — Agent Elo
 RL conquers new territory: OpenAI Five improves even further
AirStreetCapital · case_study
RL conquers new territory: OpenAI Five improves even further
99.4% — win rate
 RL conquers new territory: OpenAI Five improves even further
AirStreetCapital · case_study
RL conquers new territory: OpenAI Five improves even further
800 — Trueskill
 What's next in RL: Play-driven learning for robots
AirStreetCapital · case_study
What's next in RL: Play-driven learning for robots
85.5% — Success rate
 What's next in RL: Learning dexterity using simulation and the world real
AirStreetCapital · case_study
What's next in RL: Learning dexterity using simulation and the world real
 What's next in RL: Curiosity-driven exploration
AirStreetCapital · other
What's next in RL: Curiosity-driven exploration
 What’s next in RL: Learning dynamics models for online planning
AirStreetCapital · case_study
What’s next in RL: Learning dynamics models for online planning
180x — Data efficiency gain
 What’s next in RL: Moving research into production environments
AirStreetCapital · case_study
What’s next in RL: Moving research into production environments
 ML for life science: AlphaFold predicts de novo 3D structure of folded proteins
AirStreetCapital · case_study
ML for life science: AlphaFold predicts de novo 3D structure of folded proteins
 A big year in natural language processing: Pretrained language models
AirStreetCapital · industry_trends
A big year in natural language processing: Pretrained language models
 A big year in natural language processing: Pretrained language models
AirStreetCapital · case_study
A big year in natural language processing: Pretrained language models
 A big year in natural language processing: Researchers start sniffing GLUE
AirStreetCapital · industry_trends
A big year in natural language processing: Researchers start sniffing GLUE
88 — GLUE benchmark score
 A big year in natural language processing: Machine translation without bitexts
AirStreetCapital · case_study
A big year in natural language processing: Machine translation without bitexts
27.1 — BLEU score
 Endowing common sense reasoning to natural language models: Is text really enough?
AirStreetCapital · case_study
Endowing common sense reasoning to natural language models: Is text really enough?
 A growing interest in federated learning (FL) for real-world products
AirStreetCapital · process_diagram
A growing interest in federated learning (FL) for real-world products
 Increasing emphasis on data privacy and protecting deployed ML systems from attacks
AirStreetCapital · industry_trends
Increasing emphasis on data privacy and protecting deployed ML systems from attacks
 Expert-level diagnosis and treatment referral suggestions is achieved using a two-stage deep learning approach.
AirStreetCapital · case_study
Expert-level diagnosis and treatment referral suggestions is achieved using a two-stage deep learning approach.
 Cardiologist-level performance is demonstrated using end-to-end deep learning trained on 54k patients.
AirStreetCapital · case_study
Cardiologist-level performance is demonstrated using end-to-end deep learning trained on 54k patients.
0.97 — ROC AUC
 Deep learning in medicine: The bigger the dataset, the better the model?
AirStreetCapital · analyze_data
Deep learning in medicine: The bigger the dataset, the better the model?
0.97 — AUC
 Deep learning in medicine: Neural networks decode your thoughts from brain waves
AirStreetCapital · case_study
Deep learning in medicine: Neural networks decode your thoughts from brain waves
65% — speech intelligibility improvement
 Deep learning in medicine: Neural networks can restore limb control for the disabled
AirStreetCapital · case_study
Deep learning in medicine: Neural networks can restore limb control for the disabled
5.67% — Failure rate
 Machines learn how to synthesise chemical molecules
AirStreetCapital · case_study
Machines learn how to synthesise chemical molecules
80% — Solved ratio
 AutoML: Evolutionary algorithms for neural network architecture and hyperparameters
AirStreetCapital · case_study
AutoML: Evolutionary algorithms for neural network architecture and hyperparameters
84.3 — Test AUROC (%)
 The pace of CNN-based automated architecture search is accelerating: Facebook ups ante vs. Google.
AirStreetCapital · industry_trends
The pace of CNN-based automated architecture search is accelerating: Facebook ups ante vs. Google.
76ms — CPU Latency
 State of the art in GANs continues to evolve: From grainy to GANGsta
AirStreetCapital · industry_trends
State of the art in GANs continues to evolve: From grainy to GANGsta
 State of the art in GANs continues to evolve: From faces to (small) full-body synthesis
AirStreetCapital · case_study
State of the art in GANs continues to evolve: From faces to (small) full-body synthesis
 After image and video manipulation comes realistic speech synthesis
AirStreetCapital · case_study
After image and video manipulation comes realistic speech synthesis
 Learning the 3D shape of objects from a single image
AirStreetCapital · case_study
Learning the 3D shape of objects from a single image
 Analysis of 16,625 AI papers over 25 years shows immense growth in publication output with machine learning and reinforcement learning being the most popular topics
AirStreetCapital · industry_trends
Analysis of 16,625 AI papers over 25 years shows immense growth in publication output with machine learning and reinforcement learning being the most popular topics
16,625 — AI research papers
 Google continues its dominance at NeurIPS 2018, a premier academic AI conference
AirStreetCapital · peer_benchmark
Google continues its dominance at NeurIPS 2018, a premier academic AI conference
57 — Number of 1st author papers
 88% of 4,000 researchers who published papers at NeurIPS, ICML or ICLR were men
AirStreetCapital · peer_benchmark
88% of 4,000 researchers who published papers at NeurIPS, ICML or ICLR were men
88% — Gender distribution of AI researchers
 Compensation of senior engineers at large tech companies is approaching $1,000,000
AirStreetCapital · market_landscape
Compensation of senior engineers at large tech companies is approaching $1,000,000
$1,000,000 — Total Compensation
 At the other end of the spectrum, there’s huge growth in $1.47/hour data labelling jobs
AirStreetCapital · case_study
At the other end of the spectrum, there’s huge growth in $1.47/hour data labelling jobs
$1.47 — hourly wage
Next page →