issues
search
kdmalc
/
personalization-privacy-risk
Privacy analysis for ML and classical filtering personalization parameters
0
stars
0
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Add hold-one-out / crossval to sim results
#54
kdmalc
opened
3 months ago
0
Add smoothbatch to param.txt log
#53
kdmalc
closed
8 months ago
0
Fix batch size issues: some bs's break the pipeline for some reason
#52
kdmalc
opened
9 months ago
1
Add average performance across clients for visualizations
#51
kdmalc
opened
1 year ago
0
Change default learning rate to 0.001
#50
kdmalc
closed
1 year ago
0
Add cost comp graphs alongside cost func
#49
kdmalc
closed
1 year ago
0
Troubleshoot the difference in gradient from current code and previous that had jumps
#48
kdmalc
closed
1 year ago
0
Replace self.D with self.w everywhere
#47
kdmalc
closed
1 year ago
0
Verify that wave in gradient is due to dim_reduc
#46
kdmalc
closed
1 year ago
0
Find a better solution than prev_update/current_update
#45
kdmalc
closed
1 year ago
0
Replace current_round/current_threshold/local_round_thresh with modulus
#44
kdmalc
closed
1 year ago
0
Implement Maneeshika's Code
#43
kdmalc
opened
1 year ago
1
Understand the effect of normalizing D
#42
kdmalc
closed
1 year ago
2
Compare local decs of each client after training
#41
kdmalc
opened
1 year ago
1
Add APFL vanilla local for validation
#40
kdmalc
closed
1 year ago
1
Hyperparameter study for APFL
#39
kdmalc
closed
1 year ago
1
Refactor code so that APFL is its own class...
#38
kdmalc
closed
1 year ago
1
Doubel check self.w and D for APFL
#37
kdmalc
closed
1 year ago
0
Double check V / Vlocal / Vglobal
#36
kdmalc
closed
1 year ago
0
Implement PARS-PUSH
#35
kdmalc
closed
1 year ago
2
Set up JSON config files
#34
kdmalc
opened
1 year ago
0
Validate against various (S)GD methods
#33
kdmalc
closed
1 year ago
1
Get real hessian version of APFL to run
#32
kdmalc
closed
1 year ago
2
Break up classes into different files and into more classes
#31
kdmalc
closed
1 year ago
1
Record and plot gradient
#30
kdmalc
closed
1 year ago
0
Find L and mu according to eigs
#29
kdmalc
closed
1 year ago
2
Normalize input EMG F
#28
kdmalc
closed
1 year ago
1
Add Performance Plotting instead of just cost
#27
kdmalc
closed
1 year ago
0
Add visualization of update changes
#26
kdmalc
closed
1 year ago
2
How to correct error increase as more updates are used?
#25
kdmalc
opened
1 year ago
1
Why are the local and global error logs the exact same for the prev dec case?
#24
kdmalc
closed
1 year ago
2
Why are initial decs from CPHS all different?
#23
kdmalc
closed
1 year ago
1
Merging FL Implementation
#22
kdmalc
closed
1 year ago
0
Why is init cost so low? Random init dec should be the worst performing
#21
kdmalc
closed
1 year ago
1
SVD
#20
kdmalc
opened
1 year ago
0
Global/Client PCA Transform Model
#19
kdmalc
closed
9 months ago
1
Local_error_log bug: double print and no ID
#18
kdmalc
closed
1 year ago
0
Why does EtaGradStep blow up?
#17
kdmalc
closed
1 year ago
2
Is there a way to dynamically go to the next update?
#16
kdmalc
opened
1 year ago
1
Assess relationship between different conditions and different blocks for the same users
#15
kdmalc
opened
1 year ago
0
Is it possible to cluster users to get an even better init dec or to prevent global model divergence?
#14
kdmalc
opened
1 year ago
0
Refactor to allow local, personalized, and global error logs
#13
kdmalc
closed
1 year ago
1
When smoothbatch?
#12
kdmalc
opened
1 year ago
0
Parameter tuning for convergence / divergence of global model
#11
kdmalc
opened
1 year ago
0
Why does non-exponential updating scheme blow up?
#10
kdmalc
closed
1 year ago
2
Why do some subjects follow similar loss trajectories?
#9
kdmalc
opened
1 year ago
0
Validate if final global model from earlier runs functions well as a good init for later trials/users/block2
#8
kdmalc
closed
1 year ago
8
Check non-IID/unequal splits from ARJ
#7
kdmalc
closed
1 year ago
1
Figure out issue about no loss improvement after advancing training data update
#6
kdmalc
closed
1 year ago
1
Refactor plotting to allow direct comparison between models
#5
kdmalc
opened
1 year ago
0
Next