Schedule

This schedule is to change as the weeks go on.
Readings are to be completed before class.

Week 1

09/07

Lecture Course Introduction (note): How to Read a Paper

Week 2

09/12

Lecture File System I (note) (handout): Read S4 of UNIX Implementation
(optional) The UNIX Time-Sharing System

09/14

paper File System II (note) (handout): Rethinking File Mapping for Persistent Memory
(optional) ctFS: Replacing File Indexing with Hardware Memory Translation through Contiguous File Allocation for Persistent Memory

Week 3

09/19

paper File System III (note): Answer these questions before class

09/21

Lecture ML Systems I (note) (handout) (jupyter, engine.py): Learning representations by back-propagating errors
(Optional) Ch6.5.1-6.5.3

Week 4

09/26

Discussion Learned fs proposal (learned fs I) (handout): Lab1-challenge [Daniel and Matthew]

09/28

paper ML Systems II (note) (handout): TensorFlow: A system for large-scale machine learning
(Optional) Scaling Distributed Machine Learning with the Parameter Server

Week 5

10/03

paper Learned Index I (note): The Case for Learned Index Structures

10/05

paper Learned Index II (note): Benchmarking Learned Indexes

Week 6

10/10

no class

10/12

Discussion do we need dirs? (note) (learned fs II): Hierarchical File Systems are Dead
Lab2-challenge [Brent, Will]

Week 7

10/17

paper Learned Scheduling I (note): Learning Scheduling Algorithms for Data Processing Clusters

10/19

paper Learned Scheduling II (note) (handout) (NUMA): Machine Learning for Load Balancing in the Linux Kernel

Week 8

10/24

paper Learned Scheduling III (note): LSched: A Workload-Aware Learned Query Scheduler for Analytical Database Systems

10/26

discussion file access pattern (note) (handout): A File is Not a File: Understanding the I/O Behavior of Apple Desktop Applications

Week 9

10/31

paper Learned prefetcher (note) (handout): Predicting Memory Accesses: The Road to Compact ML-driven Prefetcher

11/02

discussion learned fs design I (note) (handout): [project updates]

Week 10

11/07

lecture Guest speaker: Andreas Kipf: Learned Cardinalities: Estimating Correlated Joins with Deep Learning

11/09

paper Caching in production (note): CacheSack: Admission Optimization for Google Datacenter Flash Caches

Week 11

11/14

dicussion learned fs design II: [project updates]

11/16

paper FUSE performance (note) (handout): To FUSE or Not to FUSE: Performance of User-Space File Systems

Week 12

11/21

paper NN4Sys explainability (note): Interpreting Deep Learning-Based Networking Systems

11/23

Thanksgiving recess

Week 13

11/28

lecture NN verification (jupyter) (engine.py): AI2: Safety and Robustness Certification of Neural Networks with Abstract Interpretation (read only FC layer and ReLU; ignore convolutional layer and max pooling; also skip proofs)

11/30

Lecture Guest speaker: Stanley Bak: Neural Network Compression of ACAS Xu Early Prototype is Unsafe: Closed-Loop Verification through Quantized State Backreachability

Week 14

12/05

discussion learned fs design III: [project updates]