Qualification for the end of round1

Woops! Apologies, missed the tag.

View post 0

Qualification for the end of round1

View post 0

GitHub repo gone?

Not sure if I’m responding in time … both repos are accessible to me at present.

View post 0

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

@hanschoi86, thanks for the added clarity. And Yikes! Out of interest, was looking at your guys “standard” vs “itemized” deductions is quite scary! Some empathy shed. What really works in the complexity is that different states may apply different taxes

Yes, the “accounting rule” I speak of applies to “businesses” (self-employment qualifies here, which is the audience I was aiming at).
I imagined that nobody would be competing as a company (you’d hence be acting as self-employed) - think being an individual or team of individuals was part of the competition rules.

Wikipedia affirms that the “accounting rule” applies to “corporations” (self-employed inclusive)

Reading in, If you’d like to pursuit this route, you need to be eligible for schedule C (or C-EZ) and file under those schedule terms. See IRS.

Found this interesting, thanks for the conversation & input

View post 0

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

@wywarren I believe they scheduled for 15 May 2019 - see this link

View post 0

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

@ross_wightman ah ok. I dispute tax law, and I encourage protesting against it. I do comply with the law non the less, doesn’t mean we can’t protest though - otherwise we’ll just continue to be taken advantage of. Gotta speak up for ourselves in today’s cruel world.

View post 0

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

@isy @ross_wightman Please bear in mind that it is your government who dictates the laws of your country, if you wanted to be upset with anyone regarding this tax dilemma, protest with your government. Tax is entirely unnecessary, and civilisation survived without income tax until less than 100 years ago. (American history, income tax was introduced to recuperate after a civil war, that civil war has long passed.)

I do agree that decoupling of the technical challenge from the taxes could have been provisioned.

View post 0

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

As mentioned above, every country’s tax is governed differently, look at your tax law, or enquire with a tax practitioner. I just qualify as a businessman.

My tax tip comes from basic accounting. Buying bread for $1 and selling it for $1, means you have 0 profit. As far as I’m aware, generally tax is applied to profit. Similarly, if you “won” $1100, and spent $1100 in the same event in trying to obtain more profit, you ultimately make 0 profit, and hence no tax implications. (unless winnings increased)

As for the length of time that the paperwork for the tax takes, it’s essentially 1 or 2 records in your ledger ? perhaps 5-10min max. Whilst this is in context of a business, it would apply similarly to a service provided by an individual. Question you’re asking is, “Did I make a profit?”. So just keep your paper trail - receipts from google cloud.

Here’s a more complicated scenario:
The entry to this competition costs $500. I ultimately win $1100 from this competition. An accountant doing their job properly would only have you taxed for $600 ($1100 - $500 ), as the entry fee was an expense in order for you to gain the ability to obtain the $1100. If you didn’t win the $1100 however, you would have generated a loss of $500 and you would not be taxed at all - but you’d be smart to keep your accounting up to date.

View post 0

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

Tax Tip: Deplete the $1100 within Round 2, then it wont be taxable - as an expense in toward generating income.

View post 1

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

Ah, interesting. I’m a team of 1, and have only seen 1 other email like that of the leader.

View post 0

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

@Miffyli I agree it may make sense that one is able to “reject” their round 1 prize. I however also see that it makes sense that all the required forms are still submitted for eligibility checking purposes - as in to make sure that you are eligible to receive a any prizes in the first place.

On the other hand, I’m of the view that not being eligible for a prize due to tax prohibitions should not subject the competitor to not being allowed to compete in following rounds, that seems a bit senseless (as if tax hadn’t disabled/disadvantaged anyone enough). I’d strongly agree with the idea of round 2 eligibility only be based on the competitive ability of the competitor, and if for any other reason beyond their competitive ability they could not receive the prize, it would be forfeited and handed down to the next leading competitor. Nothing wrong with that ?

View post 0

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

Just for clarity - and I stand to be corrected-, the FW8BEN form to be completed is only dealing with international tax, which I’ve read to be at 30% by default. Hence, we only receive 70% of the prize if the default FW8BEN is applied (look at section 9 and 10 of the form). And so if you have a default FW8BEN configuration, you can be taxed within your own country on only 70% of the prize value (as that’s what you will officially receive). (please do correct me if I’m incorrect)

View post 0

Question about the “Round 1 of Unity’s Obstacle Tower Challenge” email

@ipv6 Please understand that tax is quite complicated (especially international taxes).
The tax forms (fw8ben) is required for them to release your prize to you, and this form has very little to do with your national tax. Understanding this may clear up many questions you’ve raised. That said, you will need to consult with a tax practitioner from your country to understand whether tax needs to be paid on the prize or not. As an example, here in South Africa, if we are granted a prize that is in relation to our field of work, SARS (the equivalent of IRS in South Africa) expect us to declare the value of the prize as part of our income tax. So if you work in the field of machine learning research, it is likely that you have to declare this within South Africa. On the other hand, if you generally don’t do this kind of work, it would not be considered income tax. (I am guessing the reasons for this is to tax sportsmen where their income usually consists of competition money. It’s quite unfortunate.)

1 + 2 have been answered above.

3.) Will have to await a competition official’s response ( @arthurj )
4.) I’m also keen to know about this, our identity documents are never tightly coupled to our domicilium
4.b) The email is directed at your team leader only
5.) I’ve submitted my documents on Friday - perhaps too late for a response before the weekend. And I’ve not yet received any response yet today. I’d say this warrants more time to be granted under reasonable circumstances.

Again, I’m not an official, just a competitor looking for fair competition procedures to be followed.

View post 0

Let's talk sparse / dense rewards

@arthurj catching up on my understanding of the problems that have been observed in sparse reward systems, the objective of this competition becomes clearer to me. I was initially pulled into this competition thinking from a reward engineering perspective - classical RL. I realise I may have been mistaken.

I would like to have clarity on technical limits within this competition. There are known solutions (or solution attempts) for generalising in sparse reward systems where by the agents generate their own rewards. It’s clear to me now that these kinds of algorithms are more sought after in this competition event, however, does this mean that manually engineering rewards for OT are not part of the allowed solutions? Or is it expected that the agent only self learns particular goals/rewards - for example previously I mentioned imitation learning as a possibility for the AI to learn, and although the agent may learn to generalise on the OT, it may not be a generic solution that can be applied to other problems (each environment would require imitation - and who knows, the top of this tower may not be reachable by humans (looking forward to seeing round 2))

A bit more thought in context of your robot, I’d essentially try to cut down on training time by breaking the back-flip into smaller rewardable steps for the agent, perhaps best thought of as a form of curriculum learning. First learn to jump high enough, and get a grip on gravity, and learn to rotate in a particular direction. This is assuming the robot is equipped with an accelerometer - as I doubt the robot would ever reach a back-flip goal in a reasonable/feasible time with only CV - I’m not saying it can’t happen, but the amount of effort/resource cost doesn’t appeal as an effective solution. Hence, I’d definitely exploit an accelerometer - and attach one if it didn’t have one. I mean, there is a lot I could do with an accelerometer, and our world consists more than just a visual perspective - we have access to physics and other interesting perceptions, and creating new ways of perception is all part of the challenge in most learning (including CV).

View post 0

Libcuda.so.1 not available

Glad you got sorted!

View post 0

Let's talk sparse / dense rewards

Yes, thanks a lot for your insights. The idea around robots not having availability of a dense reward function like in the back-flip example is particularly interesting.

View post 0

Libcuda.so.1 not available

Likely your docker environment. Are you certain it is installed in docker correctly?

Have you used docker run locally?

View post 0

Let's talk sparse / dense rewards

Hi Community

I’ve been contemplating the idea around sparse / dense rewards, and have been reading into @arthurj’s OT paper. I suppose the paper inspired my curiosity more than my agent’s so far… (ha ha)

Blockquote
Reward Function The Obstacle Tower is designed to be a sparse-reward environment. The environment comes with two possible configurations: a sparse and dense reward configuration. In the sparse reward configuration, a positive reward of +1 is provided only upon the agent reaching the exit stairs of a floor of the tower. In the dense reward version a positive reward of +0.1 is provided for opening doors, solving puzzles, or picking up keys. In many cases even the dense reward version of the Obstacle Tower will likely resemble the sparsity seen in previously sparse rewarding benchmarks, such as Montezuma’s Revenge (Bellemare et al. 2013). Given the sparse-reward nature of this task, we encourage researchers to develop novel intrinsic reward-based systems, such as curiosity (Pathak et al. 2017), empowerment (Mohamed and Rezende 2015), or other signals to augment the external reward signal provided by the environment.

I am wondering why this benchmark is aimed at being a sparsely rewarding environment. I ask this as my understanding right now is more that the sparse vs dense reward is actually more of a trade-off for learning time, and only that really. Sometimes we have the luxury to implement dense rewards that cut down the learning time, other times not.
I would actually find the reward engineering quite challenging to solve some of these tasks mentioned in the paper. Or rather, the articulation (implementation) of such a reward policy. I’ve been looking into RL for a number of years now, but I can’t say that I’ve been closely tied with the community up until this challenge, so I may be missing a bit on the intention as to why the tower has initially been setup in the direction of sparse rewards.
Could the tower not have been setup without a reward function, or rather that the reward function could be seen as more of a default?

When measuring the performance between agents, yes it makes sense that they can be measured on the same reward. However, then I stumble upon the thoughts in context of this competition, are we allowed to augment the reward function that our agent/neural net receives. As I understand it, just about anything goes - i.e I can build whatever neural net ensemble taking in any params that I feed it along with the data provided by the unity gym env. I could in fact go as far as setting up an imitation learning environment, where the rewards are not used at all. Are there any technical limits ?

View post 0

Bouncing block bug

@arthurj Awesome, thanks for the hard work!

View post 0

A machine learning lecturer

A machine learning lecturer is checking up on whether her students have been doing their reinforcement-learning homework.

ML lecturer: You sit here with a look of horror on your face and empty handed, what happened to your homework assignment?
ML student: It got eaten by the TV.
ML lecturer: Progress.

View post 4